New FDA Dietary Supplement Guidance: Unnecessarily High Requirements for Safety Factors Based on Animal Studies

December 2, 2016 - 16 minutes read

The FDA’s new guidance document on new dietary ingredient notifications (NDINs) and related issues (1) contains answers to several questions, one of which is “what safety factors should be used for determining adequate daily intake (ADI) levels of the dietary ingredient if only animal studies are available?” The answer to this question may be surprising to most in the dietary supplement and food industry, as the revised draft guidance document recommends far more rigorous preclinical studies be conducted than what is typically accepted by FDA’s Center for Food Safety and Applied Nutrition (CFSAN) as sufficient when evaluating notifications of Generally Recognized As Safe (GRAS) conclusions for ingredients to be added to food.

For food additives and, by extension, GRAS substances, FDA has defined “safe” as “a reasonable certainty in the minds of competent scientists that the substance is not harmful under the intended conditions of use” (2). The safety standard for dietary supplements, in contrast, is that the dietary ingredient when used under the conditions recommended or suggested in the labeling of the dietary supplement. “will reasonably be expected to be safe.” A “reasonable expectation” standard is materially less than a “reasonable certainty” standard.  To determine the safety of food ingredients and dietary supplement ingredients (as well as drugs), studies are often conducted in laboratory animals at levels that are typically many times greater than the anticipated amount that will be expected to be consumed on a daily basis. The difference between the level of expected daily intake by humans and the level of intake by laboratory animals that did not have an adverse effect on the health of the animal (i.e., the no observed adverse effect level, or NOAEL) can be thought of as a “margin of safety” (MOS).

According to the NDIN guidance document, the MOS that should be used depends on whether the NOAEL is derived from a chronic or a subchronic study, as well as the number of studies that have been conducted.  The guidance states that if the NOAEL is from a chronic study (one year duration or longer) in animals, the safety factor to be used is typically 100.  If a subchronic study (typically 3 – 6 months in duration) is used for the extrapolation, an additional safety factor is warranted. If data from two subchronic studies is available, with a least one of them being performed in a non-rodent species, an additional safety factor of 10 should be introduced, for a combined safety factor of 1000.  It is assumed, but not stated, that the NOAEL used for the risk assessment should be the NOAEL for the most sensitive species. FDA states in the guidance that, if only a single subchronic rodent study has been performed, a safety factor of 2000 should be applied to the NOAEL if “there is no toxicity to the rodents at the maximum tolerated dose.”[*1]

The FDA qualifies its recommendations by stating that the safety factors are “approximate values, which can vary with the specific data that are available”. A greater safety factor may be appropriate if toxicity is severe or variation in human sensitivity is expected to be great, and a smaller safety factor may be appropriate if subchronic studies in rodents and non-rodents showed no adverse effects.

The FDA accepts that GRAS ingredients can be used in dietary supplements (as excipients) or as dietary ingredients; therefore the same standard for safety factors that FDA accepts for ingredients that are notified as GRAS should, in theory, be applicable to use of substances in dietary supplements, keeping in mind that the food ingredient safety standard is the higher threshold of “reasonable certainty”, compared to the “reasonable expectation” standard for dietary supplement ingredients. A review of the substances that were successfully notified to FDA as GRAS in 2015 by scientific procedures (N = 45) and received a “no objection” letter reveals that 20 of the conclusions of GRAS status relied on data from subchronic studies in rats (and no second species) to provide evidence of safety (3). Several of these 20 substances were enzymes that were used at low levels in food and readily met the NDIN requirement of a margin of safety (MOS) of 2000 between the subchronic NOAEL in rats and the estimated daily intake (EDI), however, other substances (including some enzymes) had a MOS less than 2000 between the subchronic NOAEL in rats and the estimated daily intake (EDI), such as cellulose from Penicillium funiculosum (MOS = 639), lactase from Bifidobacterium bifidum produced in Bacillus subtilis (MOS= 431), xylanase from Trichoderma reesei (MOS =412), algal oil from Schizochytrium sp. (MOS = 40) and rice hull fiber (MOS = 8.1). The latter two ingredients are macroingredients with recognized limitations in dosing in animal studies to less than 15% of the diet (approximately 7500 mg/kg bw/day in rats) due to the ability of these substances to affect nutrient balance (4).

Although the FDA guidance states that “a lower safety factor may be appropriate if subchronic studies in rodents and non-rodents showed no adverse effects”, it does not mention whether a lower safety factor would be acceptable if the only subchronic studies were performed in rats. Because large amounts of fiber would be expected to cause the same effects in non-rodents as rodents, it would be a waste of animal resources if studies on fiber (or other macroingredients) were required in a second species to be able to justify use of a safety factor lower than 2000. Regardless of the extent of historical consumption of an isolated component of food (e.g., if the dietary ingredient is something as well-studied as an isolated fiber, protein, or antioxidant from a food source), the guidance does not allow for a reduction in the type and extent of safety studies that need to be conducted; the guidance primarily focuses on the intended demographic (adults, children or women of childbearing age) and if the ingredient is to be consumed on a chronic basis, or only intermittently to determine the types and numbers of safety studies that need to be provided to show safety of the ingredient. No such focus is made for food ingredients, as food ingredients are expected to be consumed by all adults and children above the age of two.

Of the 25 other substances that were notified as GRAS in 2015, two (Luo Han Guo fruit extract and phosphatidylserine from sunflower) relied on safety data from subchronic studies in rodents and non-rodents, meeting the criteria for a safety factor of 1000 for extrapolations mentioned above.  Clinical studies also were performed with phosphatidyl serine from sunflower; however, the longest duration was 6 months.  Although the guidance states that “if human data from chronic toxicity or ADME (absorption, distribution, metabolism, and excretion) studies (typically one year in duration)[*2] are available, a safety factor lower than 100 may be appropriate”,  the guidance does not state whether performance of clinical studies shorter than one year has any bearing whatsoever on a recommended safety factor. It is believed by the author that FDA meant to refer to data from chronic toxicity studies conducted in animals, as opposed to human chronic toxicity data, but this still would not address the usefulness of clinical studies that did not evaluate tolerability. Alternatively, FDA meant to discuss human tolerability studies, and not chronic toxicity studies. Therefore, it is reasonable to assume that a safety factor of 1000 would be required from the studies that were performed in order for Luo Han Guo fruit extract and phosphatidylserine from sunflower to be used in dietary supplements. The safety factors for these substances (415 and 200, respectively) were adequate for them to receive a “no objection” letter from FDA when the intended use was notified as GRAS but are less than 1000, effectively eliminating them from consideration as dietary supplement ingredients if only an NDIN was sought.

Of the remaining 23 substances, 12 included chronic toxicity studies in rodents and/or non-rodents, as well as subchronic studies in rodents, satisfying the NDI guidance criterion for a safety factor of 100. Subchronic studies in non-rodents and/or human clinical and/or pharmacokinetic data were included for some of these substances.  The safety factors for ten of these substances were greater than 100, satisfying the NDI criterion. Safety factors for two of the three remaining substances in this category could not be calculated due to lack of determination of an EDI in the GRAS, and the safety factor for the third substance (lecithin from canola) was 11.25, considerably lower than 100. The testing paradigm for lecithin from canola included subchronic toxicity studies in both rodents and non-rodents, plus a chronic study in rodents, but these still would not pass the NDI criterion for use of a safety factor of less than 100, because human clinical or pharmacokinetic studies were not conducted for the material.

The majority of the remaining 11 substances that were notified as GRAS in 2015 relied on data from subchronic rodent studies (some of which were less than 90 days in duration), in addition to short term clinical or pharmacokinetic studies in humans.  Seven of these substances had safety factors considerably under 100, yet they were notified as GRAS. Like several other substances that were notified as GRAS in 2015, these GRAS substances that received “no objection” letters from FDA on their intended uses in food would not meet the strict testing requirements of new dietary ingredients stated in the draft guidance if the data set had been submitted to the FDA as an NDIN.

The overriding conclusion from this analysis is that the evaluation of the safety of ingredients under draft NDIN guidance requires a greater number of preclinical studies, as well as human tolerability or pharmacokinetic studies in many instances, compared to food ingredients that also have gone through a notification process with the FDA. The number and complexity of animal studies stated in the guidance for dietary ingredients makes it highly unlikely that any ingredient other than an enzyme could achieve the types of safety factors the FDA wants to see for NDIs without performing long term and unnecessary tests in animals and humans, making the safety testing for dietary ingredients cost-and time-prohibitive, essentially blocking the ability of dietary supplement companies to innovate with new ingredients. While the author is an advocate for thorough testing of ingredients added to food and dietary supplements, there needs to be some recognition of the characteristics of the ingredient that could impact the testing paradigm for dietary supplements and reduce redundant or unnecessary preclinical studies, like there is for food ingredients.


[*1] The author believes the use of the term “maximum tolerated dose” (MTD) is in error, and rather, the term “maximum dose” should have been used instead, because the MTD is “not a nontoxic dose and is expected to produce some level of acceptable toxicity by the chemical” (5). Furthermore, the FDA provides no guidance about the safety factor to use if a single rat study is employed and there is toxicity at the maximum dose. As the guidance is a draft document, it is hoped that the issue with inappropriate use of MTD will be corrected in the finalized version of the document.

[*2] As pharmacokinetic studies are typically short term and cannot be conducted over a year with any practicality, it is assumed that the FDA misplaced the comment about a study length of one year in the draft document and this qualifier applies to the duration of the chronic toxicity study only.



  1. FDA (2016). Dietary Supplements: New Dietary Ingredient Notifications and Related Issues: Guidance for Industry, available at
  2. 21 CFR 170.3 (i).
  3. FDA (2016). Substances Generally Recognized as Safe Final Rule, available at
  4. Borzelleca, J.F. (1992). Macronutrient Substitutes: Safety Evaluation. Reg. Tox. Pharm. 16: 253-264.
  5. Hayes, A. Wallace, editor. Principles and Methods of Toxicology. 4th Philadelphia: Taylor & Frances, USA; 2001

Tags: , , , ,