A key test purporting to prove the safety of virtually all deodorant soaps is flawed by "a number of irregularities" and is "not scientifically acceptable," Food and Drug Administration officials said yesterday.

They also expressed concern about the reliability of studies on laboratory animals that claimed to establish the safety of various medicines including an arthritis drug.

All of the questioned studies were done by Industrial Bio-Test Laboratories-(IBT which tests food additives and pesticides as well as drugs. its clients include manufacturers that submit data to the Environmental Protection Agency, the National Cancer Institute and the FDA, as well as these and other federal agencies.

IBT refused an invitation from Sen. Edward M. Kennedy (D-Mass.) to join FDA officials in testifying before his Senate Subcommittee on Health and Scientific Research. The firm is a Northbrook, III., subsidiary of Nalco Chemical Co.

The deodorant soap test involved triclocarban (TCC), a germ-numbing chemical.The FDA approved it in 1963 without requiring animal tests to find out if it could be absorbed without harmful result by persons who, over long periods, used soaps containing it.

In 1971, the Monsanto Co. maker of TCC, retained IBT to do a two-year study involving rats. Preliminary results led a panel of FDA expert advisers to conclude that the amount of TCC in soaps was far below a dangerous level.

A final report on the study didn't get to the FDA until last year. Then, acting commissioner Sherwin Gardner testified, "careful review of the raw data revealed a number of irreglariesU." One problem, said agency scientific investigator M. Adrian Gross, was the "hopeless" disarray of records.

FDA drugs chief J. Richard Crout said the data permit a conclusion neither of safety nor of hazard. If the agreent ultimately rejects the study, it will have to a repeated before the FDA can act.Meanwhile, marketing of TCC soaps will continue.

Monsanto, in a press release, defended the IBT study and said repeating it would only yield "the same findings."

Garnder and Gross compared IBT's data on TCC to its work for Syntex Laboratories on Naprosyn, a rheumatoid arthrities drug prescribed for hundreds of thousands of people since its introduction last March. Syntex also sells the drug in at least 36 other countries.

IBT conducted the only animal test purporting to establish the long-term safety of Naprosyn. Gross told the subcommittee last July that one of the "really horrible" findings was that some rats in the study - if IBT records are taken at face value - had died "several times." Most that died were allowed to decompose and consequently were "practically useless" for study purposes, he said.

Yesterday, Gross testified that there are indications of borderline statistical significance that Nsprosyn causes cancer. Crout questioned this. He emphasized that the FDA, in a legal proceeding begun last October to halt sales of the drug, contends only that the study constituted "untrue statements" in the Syntex marketing application.

Kennedy said the law "doesn't make sense" because it allows Syntex to go on selling Naprosyn despite the unreliability of the only long-term safety study that was the basis for letting it go on sale in the first place.

After other animal-test investigations last year, the FDA asked the Justice Department to make grand jury inquiries of G.D. Searle & Co. of Skokie, III., and Biometric Testing. Inc., of Whippany, N.J. The cases are pending.

On Feb. 22, Sandoz Pharmaceuticals of East Hanover, N.J., told the FDA that it had found discrepancies in a rat study IBT did for it on the drug Restoril. "Some animals died more than once and others disappeared and reappeared," the firm said in one example.

With bipartisan support last year, Kennedy got an extra $16.5 million for the FDA to monitor all animal-testing labs and to improve human testing.

He told the FDA yesterday that the agency appeared to be insufficiently aggressive. He offered to subpoena IBT records to find out the products it has tested and critized the agency for failing to ask all drug manufacturers to list products tested by IBT.