environment

Watanake heads Dow's in-house effort.

ment; or corporate pressure may have caused the lab to hasten a test at the expense of accuracy. The law remains unclear about who is required to pick up the tab to redo an experiment if it has to be aborted at some point. Nalco Chemical, IBT's owner, has set up a multimillion-dollar fund to cover suits being brought by companies claiming to have been defrauded by IBT. But no one knows yet whether any precedents set by BT suits will extend to cases where Barbee is crossing labs off Olin's list.

labs are charged with sloppiness rather than fraud.

Thus, while some companies say they are for the first time considering suing contract labs to have tests redone, others are showing themselves willing to accept questionable data when a test has gone on for a while. "Science should be the only guiding principle in a company's decision [to accept or reject test results]," says Hamm. "But when a company has already spent a couple of million dollars on a study, its criteria in accepting it tend to be less rigorous." Too amenable. Charges of placing fi nancial considerations above scientific ones fly both ways. Chemical industry specialists charge that labs will sometimes sign a contract with one company guaranteeing it disease-free animals for the life of a two-year experiment, while conducting separate tests in an adjoining room for another sponsor involving the mechanics of that same disease. The result can easily be inaccurate results for both tests. Insists one angry executive: "They should be saying no to one of these studies."

The testing industry has come a long way toward avoiding inadvertent errors. For example, it is working with a new generation of computerized measuring and record-keeping systems. While lab workers were prone to transcription errors when recording information in log books by hand, the new systems can automatically record an animal's weight.

Still, while the margin for unintentional error is slimmer, deliberate errors are easy to make. Both the Environmental Protection Agency and the Food and Drug Administration, whose investigators cracked the IBT story, have stepped up their program for periodic audits of test data. In 1980 the FDA introduced its Good Laboratory Practices program, which attempted to standardize testing protocols and reporting methods, and which has been welcomed by many labs as a way to prove their practices are good. "The system establishes a paper trail of documentation that follows an experiment from beginning to end," says Hazelton's Snyder. Less to chance. Chemical companies, too, are leaving less to chance. The industry has funded CIT's toxicology lab, which is focusing on toxicity questions' for commodity chemicals such as formaldehyde and chlorine. Such chemicals have shown to be toxic in animal tests, and CIT's experiments are trying to gauge the relevance of animal findings for humans.

CIT also has conducted seminars to

alert industry toxicologists to lab testing problems. And indeed, many chemical companies have established their own teams of auditors who review both in-house and contract lab procedures. Olin's Barbee has visited many independent laboratories himself, and his inspection tours have already resulted in five labs being crossed off Olin's list for future work.

Snags. Among such basic tasks as checking animal feed to ensure that it contains proper levels of test chemicals, auditors determine whether other experiments being done in the lab could jeopardize test results. Barbee and other inspectors try to ensure that data collection is complete enough to allow for a retrospective reconstruction of the test. "It isn't a serious problem if you happen to miss one body weight for an animal that has been weighed 100 times," says Arthur F. Uelner, Monsanto's manager of quality assurance. "But if you sacrifice an animal and start making mistakes at the necropsy, then things can get serious."

Monsanto is one of several companies to take matters into their own hands. In 1978, a year after the IBT crisis first be

The testing industry is working with new equipment to avoid writing another ghoulish tale

came known, Monsanto opened the doors of a $12 million Environmental Health Center, which now conducts some 40% of the company's toxicity testing.

Chevron, too, decided to beef up its in-house testing capabilities that year. Until then, it had employed a testing staff of a dozen, and charged them only with toxicological studies on petroleum products. Now it has 85 employees, 35 of whom are professionals in such areas as toxicology, medicine and veterinary medicine. And these employees will be tackling long-term pesticide studies as well. When the company's $17.2 million Environmental Health Center is completed some time in 1985, the staff will have at its disposal 25 animal rooms that have been set up for internal pesticide testing. "The center will enable us to exercise greater control over our scheduling and monitoring of studies," predicts J. N. Osperson, the center's vice-president and general manager.

It is that feeling of control that has made in-house facilities so popular of late. Chemical companies feel safer providing proprietary information about a substance's planned uses to their own

August 24, 1983/Chemical Week 33

environment

employees, and those employees can then set up test protocols that are targeted directly at the substance in question. Moreover, an in-house facility must operate along the priority agenda of the company that owns it. An independent laboratory can be overbooked, and experiments from different clients have to wait their turn.

Testing with Yale. The cost of in-house facilities may not always be justified by the advantages, however. Five years ago, Olin was talking with Yale University about building a cooperative testing facility to handle both straightforward safety testing required for product registration as well as academically oriented studies of chemical toxicity. The project never did take off. "We estimated that it would cost us $12 million to build the facility," says Barbee. Salaries for full-time toxicologists, amounting to an estimated $180,000 annually, came on top of that. "We were only doing between $500,000 and $1 million in testing annually [ourselves], so the cost wasn't justified," Barbee explains.

Some of the larger companies run into the exact opposite problem: Their testing needs are so extensive that their in-house facilities cannot handle the load. Du Pont has operated a testing lab since the 1930s, and Dow Chemical built its own laboratory in the 1940s, yet both companies often farm out specific tests to contract labs. Sometimes, says Philip G. Watanabe, Dow's director of toxicology, there is "simply a question of Calo says lab costs are very competitive.

Conventional wisdom has it that only one out of every 15,000-20,000 products conceived in corporate laboratories makes it through to the marketplace. Some turn out not to do what they're supposed to do; others run into unsolvable toxicity problems; still others turn out to be economically unattractive when they've been reformulated to be safe. Add up the costs of testing all the losers to those run up by an occasional winner and "you're talking $35 million to $40 million for one new product."

So says Klaus Saegebarth, director of agrichemicals research and development for Du Pont. Saegebarth should know he watched Du Pont's Glean, a herbicide for control of weeds in wheat, barley and oats, go through the testing wringer at the company's Haskell Labs, not far from Newark, Del. Glean, a sulfonylurea herbicide whose principal ingredient is 75% chlorsulfuron and which can be used in applications as low as one ounce per acre, passed the tests with flying colors and was commercialized last year. The chronology of what the compound

space in our own toxicology lab." The company has used independent labs to test materials it uses in its own research (box, p. 36). And IBT conducted toxicity tests for Dow on both 2,4-dichlorophenoxyacetic acid and picloram. A mixture of the two chemicals was employed in Vietnam as the defoliant "Agent White" and is sold domestically by Dow under the tradename Tordon.

Although costs, as Hamm charges, are certainly a factor in choosing an independent laboratory, many industry toxicologists balk at any suggestion that they weigh costs above other factors. "Costs are usually close," notes Charles Calo, chief toxicologist at Velsicol, which farms out all of its toxicity testing. "The choice is made on who will give us the best job.'

Rewarding excellence. The definition of "best job" has many facets, some of which have little to do with scientific expertise. Writing ability, for example, often enters the picture. Too often, notes Velsicol's Calo, reports are rejected by a federal agency as much for their lack of clarity as for the data they contain. Calo says he will give special preference to any laboratory that has

went through gives a good picture of how arduous and expensive testing can be.

Glean's first toxicity tests were acute, short-term experiments to obtain information on the material's inherent toxicity, to develop a baseline for further studies and, most important, to see whether the material is safe for workers to handle. Groups of white laboratory rats were exposed to multiple-dose levels to establish the ld 50 or le 50-the lethal dose or concentration that is fatal to one-half of the animal population tested. For pure chlorsulfuron, the lethal dose turned out to be 5,545 mg/kg for male rats and 6,293 mg/kg for females. "These acute tests are done purely for the protection of our own people," says John A. Gardiner, manager of agrichemicals registration and regulatory affairs at Du Pont.

The skin absorption and dermal-effect tests that followed showed that chlorsulfuron was not a skin irritant on rabbits or a sensitizer on guinea pigs. Acute testing for eye effects produced very mild temporary conjuncti

been consistently praised for the literacy of its reports by FDA. "If I know we are seeking government approval, I will consider this lab," he explains. "My job is made easier when a lab is good at reporting."

Calo and several other industry toxicologists also say that they will try whenever possible to spread their work around different labs. The reason, they claim, is fear that a heavily patronized laboratory, eager to cater to an important customer, will try too hard to pro

'The difficulty that contract labs are having with quality and timing is almost universal'

duce favorable test results-only to have those results torn apart by federal agencies in Washington.

While most independent labs bridle at the idea that they are so easily pressured, they do concede that more controls are needed. Through NALSI, they are trying to set up an effective selfpolicing system. The organization is encouraging all of its members to have their facilities inspected and approved

34 Chemical Week/August 24, 1983

28-401 O 84 15

val irritation in rabbits following the administration of 10 mg of chlorsulfuron. Also, in the acute inhalation phase of the initial tests, groups of rats were exposed to chlorsulfuron for a single four-hour period at a number of different exposure levels.

Glean sailed through these initial tests. Its next hurdle was field trials, where its efficacy was amply demonstrated. Then, the next round of toxicity tests began. "Once we start on this avenue, we have made a major decision that puts the company into major toxicological testing and expenditures," says Gardiner, who estimates that a complete battery of toxicity tests will cost a company between $4 million and $5 million.

Scientists started the second phase of testing with subchronic oral tests, which involve repetitive exposures to the material at lower levels than the initial studies. For experimental fielduse permits, for example, the Environmental Protection Agency requires 90day feeding tests on rats and dogs. Subchronic oral tests on Glean established no-observable-effect levels (NOEL) of 100 parts per million in the diet of rats and a NOEL of 2500 ppm for mice and dogs.

Those test results established the

baseline for Glean's third round of toxicity trials, which involved two-year studies of feeding Glean to test animals at three different dose levels. The highest dose level was set to show a measurable effect; the lowest level, no effect. Midpoint levels are normally set 2 to 5 times higher than the NOEL, explains Phillip W. Schneider, Jr., a Du Pont toxicologist. "We're testing for chronic toxicity

Multiple-phase scrutiny of a herbicide's safety shows the high cost of finding a winner

long-term effects of low levels in the animal body,"he says.

The highest level fed was 2,500 ppm in rats and 5,000 ppm in mice. Glean established a NOEL of 100 ppm in the diet of rats and 500 ppm in the diet of mice.

Glean turned out to have no carcinogenic effects at any level. Thus, researchers passed on to the next crucial testing hurdle: EPA-required tests on rats and rabbits for teratogenicity, or potential to cause birth defects. Chlorsulfuron came up clean for teratogenicity at levels up to 75 mg/kg in rabbits.

The Haskell lab tested three generations of animals, including those still in utero. Investigators checked for effects on animal fertility, lactation and live animal fetuses. Earlier this year, EPA proposed to reduce the number of generations and litters studied, but Gardiner says that Haskell will continue running the tests as they do now. Part of the reason is that Du Pont must meet more stringent testing requirements established by Canadian regulators when marketing products like Glean in that country.

Glean's last major series of tests were to determine mutagenicity. A recent development in toxicity testing, these short-term tests look at the action of a chemical on animal genes, and also test how quickly genes repair damage caused by the test material. Here again, Glean was given a clean bill of health, says Gardiner.

While all toxicity studies on Glean were performed in-house, company officials say that they do contract for outside work with labs that have specific expertise. As an example, DuPont's Saegebarth cites wildlife studies on the effect of such pesticides as Glean on mallard ducks and bobwhite quails. Both tests are required to meet EPA pesticide regulations.

by the Toxicology Laboratory Accreditation Board, or "T-Labs." Another outgrowth of the IBT revelations, T-Labs was formed in 1979 under the auspices of the Society of Toxicology. "It is designed to be a peer review system to monitor toxicology labs for standards of performance," says Emil Pfitzer, director of Hoffman-LaRoche's department of toxicology and pathology and chairman of T-Labs.

Selective approval. The group's 12-person accreditation board is equally balanced between scientists from academia, government, contract labs and industry. Upon receipt of an application for lab accreditation, several board members will inspect the facility. Pfitzer says the group does not give a wholesale seal of approval to a lab, but only accredits those categories of a lab's total toxicological testing program that meet T-Labs' standards.

To date, T-Labs has given full accreditation to one contract lab, Hazelton, and to corporate facilites operated by Schering and Union Carbide. Pfitzer ex

Mice in hand: Toxicologists study the effects of new chemicals on lab animals.

environment

Dow took nothing for granted

Despite the regulatory emphasis on toxicity testing, companies cannot take it for granted that the chemicals they buy have been given a toxicological all-clear. Unregistered chemicals, particularly those sold in small quantities for use in research, often enter corporate laboratories without the knowledge of the Environmental Protection Agency or other regulatory body.

Many chemical companies routinely screen chemicals supplied to them by other manufacturers and distributors. And, as Dow Chemical learned, those tests can be lifesavers.

"

The case in point involves a small quantity of N,N-dimethylaminoacetonitrile, which Dow acquired for what the company's director of health and environmental sciences, Etcyl H. Blair, will only describe as "R&D purposes. The material was purchased, Blair says, through a catalog from Fluka Chemical (Hauppauge, N. Y.), the American distributor for Fluka A. G. of Buchs, Switzerland.

Dow submitted the material to two contract laboratories to test its effect on the animals. The material turned out to be a fast-acting poison. For example, one test showed that minute

pects the NALSI program to rapidly increase those labs seeking T-Labs accreditation.

How much more pressure will come from Washington remains an unanswerable question. In 1977, following the original BT investigation, FDA and EPA inspectors swept through contract laboratories nationwide. The results were not particularly dramatic, says Edwin Johnson, director of EPA'S Office of Pesticide Programs (OPP). "The thing about BT was that there was a pattern of misreporting that ran through almost all of its tests," Johnson says. "We saw problems at other labs but they were related to specific studies."

Easygoing. There is a growing feeling in Congress, however, that EPA has been too soft on the labs. Indeed, a December 1982 House Agriculture subcommittee report charges that EPA's auditing system, established in 1977 as part of the OPP, has thus far been a wash. No "decisive regulatory or enforcement actions have been taken as a result of [OPP's] laboratory audit program," the report states.

36 Chemical Week/August 24, 1983

but lethal doses of N,N-dimethylaminoacetonitrile were readily absorbed through a rabbit's skin with minor traces of surface irritation.

Blair immediately contacted EPA's Office of Toxic Substances to alert of ficials that the chemical's "true toxic potential" had not been revealed in

Published toxicity data

on a catalog chemical failed to fully identify its hazards

previously published toxicity data. "The fact that lethal doses could be absorbed so readily is what made us concerned," says Blair.

EPA receives a number of similar "for your information" letters annually, says Frank Kover, chief of EPA's Chemical Hazard Identification Branch. If the letters deal with chemicals already in commerce, EPA can require manufacturers to produce more toxicity data, or can, as a last resort, pull the product off the market.

For unregistered or experimental chemicals, however, EPA can do little more than sound an alarm to the chemical's manufacturer and those

The FDA, by contrast, has earned a reputation for extreme vigilance. At a July 27 House Agriculture subcommittee hearing, subcommittee chairman George E. Brown, Jr. (D., Calif.) laid out the differences. Noting that both agencies had received approximately the same number of long-term cancer studies last year, Brown said that statistics compiled by his subcommittee showed that while FDA Scientists spent some 100 man-days in reviewing each study, EPA staffers breezed through the same assignment in 10 man-days. Similarly while FDA utilizes 20 persons in its lab audit program, a similar unit at EPA employs two. Last year, Brown added, "not a single long-term [toxicity] study on a pesticide" had been audited by EPA. "It is increasingly clear," Brown says, "that EPA lacks the necessary resources and expertise to adequately accomplish these basic scientific tasks on a routine basis."

Not surprisingly, EPA thinks the criticism is too harsh. "I'm not going to tell you that we go out and audit enough labs, but the program is not quite as

who might come in contact with the substance. As for regulating imports of such substances, Kover says that regulations are currently pending that would require an importer annually bringing in more 10,000 kilos of an unregistered chemical to report toxicity data on the substance to EPA. He acknowledges that the rule will do nothing to prevent smaller quantities of highly toxic chemicals from entering the country.

It is not always an easy matter to trace such chemicals back to their source. In the case of the N,N-dimethylaminoacetonitrile, Paul Kuehnis, general manager of Fluka A. G., says that his company does not produce the material itself. Asked by a reporter to name the company that had supplied the highly toxic chemical to Fluka, Kuehnis declined, citing a matter of "principle.'

Kuehnis insists that Fluka sells only the "smallest quantities" of such chemicals through its catalog. And he contends that the firm's customers are "experienced chemists" who know how to handle highly toxic substances. However, Kuehnis concedes that his company does not attempt to check this out. Citing another matter of "principle," he says, "our customers do not want us to know what they are doing."

limited as Mr. Brown's comment makes it seem," says OPP's Johnson. He contends that EPA does much of its work through FDA inspectors, and that when one adds up the slices of those people who contribute in some way to the EPA's auditing program, the total effort equals 13 to 15 full-time persons."

Moreover, Johnson insists that EPA has moved swiftly when it deemed fit. Because of an EPA audit, St. Louis-based American Histolabs conceded that it had botched some slides of tissue samples it was asked to analyze for FMC, and is redoing the tests. In another instance, EPA scrutiny of a Florida lab, Chemiehaus, led to the closing of that operation. "It was a one man operation and the person was totally unqualified"says James V. Rolofs, an OPP staff assistant. "He had some weird degree from a Yugoslavian university."

EPA has also prosecuted one case that the Justice Dept. has declined to press. The agency filed civil charges against McKwin-McConnon, a Minnosota-based lab. According to Rolofs, the lab had submitted identical test data for two