difficult for FDA to be clearer on how to demonstrate strain specificity. As an illustration, he described the difficulty microbiologists have in clearly differentiating pathogenic versus nonpathogenic strains of the same species, such as Escherichia coli.

Just as there is no scientific consensus on how to distinguish among strains, there is no scientific consensus on how best to evaluate the safety of a microbial culture. This is unlike nonmicrobial ingredients, for which the guidance provides references to standard toxicology assays. “There are protocols out there that tell one exactly how to do each of those studies,” Levy said. Not so for live microbes. As with identity, the guidance provides only general advice about what information should be submitted to demonstrate safety, since this is a rapidly changing area of science.

Research on the microbiome is such a rapidly advancing field that it is premature for FDA to develop specific recommendations at this time, Levy remarked. He referred to the same Agency for Healthcare Research and Quality (AHRQ) study on probiotic safety that Mary Ellen Sanders had mentioned previously (Hempel et al., 2011). The authors of that study concluded that while there is no evidence that probiotics are unsafe, they did not have a great deal of confidence in their conclusion. The review found that most trials reported in the scientific literature are either poorly designed or poorly reported, making it difficult to evaluate safety. Many do not mention which strain was used or how the strain was prepared. Others do not explain how adverse events were monitored. The review also found that there has been no real effort to examine long-term safety risks (either safety of long-term exposure or long-term effects that show up after the conclusion of a trial). “Nobody is studying long-term safety in a systematic way,” Levy said.

Because the guidance document cannot be specific about areas in which science is rapidly developing, Levy encouraged companies to engage in dialogue with FDA before submitting NDI notifications. “We are using these organisms in new way,” he said. “[The] organisms … sound familiar because they have been present in fermented foods for a long time, but we are selecting them to have properties that were not really selected for previously. That is an intended use that is new and requires a dramatic new efficacy and safety evaluation paradigm.”


Levy’s presentation prompted several questions on evaluating probiotic safety. A workshop audience member asked if FDA’s “typical approach” of using an existing benchmark and examining marginal changes from that benchmark could be applicable in the case of live organisms. For example, if an organism already exists in a yogurt product, would it be acceptable

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement