ORAFLEX MAY HAVE COMPASSIONATE USE, FDA ACTING COM. NOVITCH TELLS REP. WEISS; FDA REVISING GUIDELINES TO REQUIRE TWO-SPECIES CARCINOGENICITY STUDIES FOR NDA APPROVAL

FDA does "not wholly agree" with Lilly's decision to suspend all investigational uses of Oraflex based on the firm's mouse carcinogenicity studies, Acting Com. Novitch told Rep. Weiss (D-NY) in an April 9 letter regarding the role of animal studies in drug approval decisions. "We do not wholly agree with Lilly's decision to terminate, at this time, all investigational uses of Oraflex" (benoxaprofen), Novitch said. "The animal findings, at this point, certainly appear convincing enough to terminate all formal studies," he explained, but the decision to halt "all 'compassionate' uses should be reconsidered in light to the new findings." Novitch said "We should not rule out the possible use of Oraflex in a patient who has exhausted all available alternatives and who is made fully aware of the animal data," Novitch wrote, adding: "We would expect such uses to be infrequent." Novitch was responding to a Dec. 23 letter, in which Weiss queried the agency about its policy on the number of carcinogenicity studies required for NDA approval. The House Intergovernmental Relations Subcmte. chairman's letter followed by two days Lilly's decision to terminate all investigational uses of benoxaprofen due to findings of liver cancer in mice ("The Pink Sheet" Jan. 9, p. 7). Weiss had noted that guidelines by PMA, the Natl. Cancer Institute, and the Natl. Toxicology Program all recommend carcinogenicity studies in two animal species, but FDA requires only one animal study. The agency reviewed a rat study, which showed no major carcinogenic effects of Oraflex, before the antiarthritic was approved in 1982; a year earlier, as a condition of marketing, Lilly began the mouse study, which was completed in April 1983. FDA will revise its toxicology testing guidelines "in the near future" to require two valid carcinogenicity studies in different animal species as a prerequisite for approving new drugs, Novitch told Weiss. "Although our de facto requirements, as well as drug company adherence to PMA guidelines developed in 1977 (reflecting joint FDA-PMA activity), have assured adequate carcinogenicity testing in nearly all cases, I believe our guidelines should be changed formally to reflect the current consensus that studies in two species are usually needed for an adequate carcinogenicity assessment," Novitch asserted. "We therefore will propose a new guideline for chronic toxicity testing in the near future," he said, adding that the two-species standard "will ordinarily be required for drugs with chronic or intermittent use." To illustrate the agency's de facto pre-approval standard of two carcinogenicity studies, Novitch cited the 16 new chemical entities (excluding anticancer drugs) which were approved by FDA in 1982 and 1983 for chronic or repeated intermittent use. "It can be seen that in all but five cases, two rodent species were studied prior to approval," the acting commissioner said. The five drugs are Oraflex, Roche's Bumex (bumetanide) and Accutane (isotretinoin), Wyeth's Wytensin (guanabenz), and an orphan drug for infected stone diseases, acetohydroxamic acid, whose labeling warns of likely carcinogenicity. Furthermore, Novitch said, a number of drugs used in short-term regimens -- including Purdue Frederick's Prioderm (malathion), Burroughs-Wellcome's Zovirax (acyclovir), and Miles' Niclocide (niclosamide) -- also were studied in two species for carcinogenicity prior to approval. When Oraflex was approved, FDA had "received a mutagenicity battery of studies, all of which were negative," the acting commissioner pointed out. "There was no indication that the drug had any carcinogenic potential." A single long-term rat study "was considered to have fulfilled our then-current carcinogenicity study guidelines," he added. In addition, Novitch continued, "interpretation of the results of carcinogenicity studies is not always as straightforward as a mere application of a standardized statistical test, but is rather a sophisticated evaluation problem" insofar as "many comparisons of a treatment and control group are made for many tissue sites." The decision process regarding the potential toxic effects of a test chemical "should incorporate biological knowledge and corroborative evidence, such as the presence of a dose-response relationship or an elevated tumor incidence in more than one sex-species, and consideration of other evidence, such as structure activity relations, mutagenicity, genetic toxicology, and chemical disposition," he added.