On Friday, June 4, the FDA Center for Veterinary Medicine (CVM) published an announcement regarding a Generally Recognized as Safe (GRAS) Notification program (Substances Generally Recognized as Safe Added to Food for Animals; Notice of Pilot Program). However, while the announcement is a good start and pledges fidelity to the Center for Food Safety and Applied Nutrition’s (CFSAN) 1997 “Substances Generally Recognized as Safe; Proposed Rule”, there are some inconsistencies with the CFSAN Proposed Rule and some issues left unaddressed that have been foci of concern for both CVM and the stakeholders.
Early in the announcement, there appears to be an incongruity regarding eligibility for participation, in that at the header titled “DATES” the following language is used: “Submit written requests to participate in the pilot program”. However, the first line in Section III of the announcement indicates that “Any person may notify FDA of a claim [for GRAS exemption]…” Presumably, this is not an inconsistency, but an encouragement to prospective Notifiers for a pre-Notification meeting with CVM. However, does this pre-Notification meeting represent an opportunity for CVM to head off a potential Notification of a substance CVM has decided embodies too much “risk”? In the past (as recent as September, 2009), FDA personnel described GRAS as a pathway for substances of low priority for regulation (much akin to those currently eligible for “regulatory discretion” letters). Does this mean that those substances with potentially higher “risk” would be forced into the food additive petition queue? Presumably, heading a GRAS determination off at the pass before it is declared GRAS and subsequently Notified, would prevent a potentially embarrassing high rejection rate as is seen with CFSAN’s New Dietary Ingredient Notification program (~70% rejection rate). But, such a pre-GRAS determination selection process infringes on the entitlement of any manufacturer to determine the statutory route by which his substance may become compliant i.e., GRAS or Food Additive Petition. Judgment of a Notification or validity of a GRAS determination solely on the basis of CVM’s determination of potential “risk” is not a call CVM is empowered to make; this is a GRAS Notification process (proposed 21CFR170.36), not a GRAS Affirmation process (21CFR170.35 or 570.35).
The announcement also engages in some redefining the working practices at CFSAN in section I.B. “Elements of the GRAS Standard”. First, the announcement declares that “…use of a substance that is GRAS requires both technical evidence of safety and a basis to conclude that this technical element of safety is generally recognized.” The “technical element” refers to the rigor of test data supporting the determination and the cases cited are generally rooted in the Weinberger v. Hynson et al. decision, wherein the “substantial evidence” doctrine is applied:
However, when referring to Weinberger v. Hynson et al., CVM must realize this is not a license for the Agency to solely determine what constitutes “substantial evidence,” as “substantial evidence” is (according to the decision) determined by “experts qualified by scientific training and experience” and not necessarily only those experts working at CVM. As if to ensure primacy over any expert panel in a determination of safety, the following statement is also made:
(2) there must be a basis to conclude that there is consensus among qualified experts about the safety of the substance for its intended use.
That is, this statement makes it clear there must be a “consensus” regarding safety and that an independent expert panel’s opinion may not represent a consensus. Further, this statement is followed by citations of cases in which FDA’s assertion prevailed that there was not a consensus among experts. Together, these statements and citations indicate that the Agency will reserve the right to overrule the finding of GRAS by an independent panel of experts on the basis that a “consensus” was not achieved. To reinforce this assertion, a subsequent section of the announcement discusses the term “consensus” as follows:
In this document, FDA is using the term “consensus”’ in discussing the common knowledge element of the GRAS standard. Such consensus does not require unanimity among qualified experts.
Ordinarily, the reader would expect that this statement would be a defense of the panel of “experts qualified by scientific training and experience” as representing mainstream scientific thought, although acknowledging the possibility of a minority opinion. However, in the cases cited (US v. Articles of Drug***5,906 Boxes, et al.), again all cases are those in which FDA prevailed in its claim that a substance was not GRAS. Therefore, this statement is a tad counter-intuitive as it does not reinforce an expert panel in its ability to voice a consensus and reject a minority opinion, but just the opposite, as it illustrates circumstances in which an expert panel opinion did not constitute a “consensus” and was refuted by the Agency.
Other items which may represent a deviation from the CFSAN proposed rule, include:
Allowing publication of data in the “secondary scientific literature such as review articles, textbooks and compendia” – CFSAN has historically shied away from “secondary literature” preferring to see original papers. However, while CFSAN may see this as an issue related to the “common knowledge threshold”, CVM may regard secondary references as evidence of consensus among experts.
A requirement(?) for publication of supporting science in “peer-reviewed scientific journal(s)” – a requirement not demanded by CFSAN, the latter of whom has allowed publication in non-peer reviewed journals or disclosure of reports on a private website. However, in light of the above, CVM may have a different purpose with the demand for peer-reviewed publication, as this could be argued as meeting both the “common knowledge threshold” as well as “evidence of consensus” if the peer reviewers represent a “consensus”. Interestingly, the announcement indicates “documentation of the opinion of an ‘expert panel’ that is specifically convened for this purpose” may partially suffice as meeting the common knowledge element, but could arguably be in pursuit of “expert consensus” as well. Does this imply that publication of the expert panel consensus statement would be adequate to meet the consensus doctrine discussed above and, if so, does this circumvent the peer-reviewed journal requirement?
In sum, it appears that “consensus” has assumed primacy over the “common knowledge” element for a GRAS determination.
Not fully discussed in the article were the more frequently discussed points, including the following:
Utility. CVM has historically promoted a very narrow definition (near Procrustean for food source animals) of utility that divides all food ingredients into those providing “nutrition, aroma and taste” and, all others as drugs. A few exceptions included feed enzymes, which may work in the gut of the ruminant to make nutrition more available. Such a strict division of uses overlooks all the “technical function” substances provided in 21 CFR 170.3(o) such as preservatives, texturizers, buffering agents, etc. However, because it appears that CVM may be abandoning the long held position that even probiotics were drugs, it could be that CVM is reconsidering how it defines “utility.”
Target species. Nothing more specific on this issue in the announcement, other than the statement that target species data must be submitted. The possible contentious issue here is the quality and quantity of safety data specifically developed in target species versus how much extrapolation from non-related species is allowed or what sort of bridging data would be satisfactory. To be fair, this will have to be worked out on a case-by-case basis, but a mention of acceptance of non-target species data would apply a measure of comfort, especially among companion animal food manufacturers.
Time frame. The announcement indicates “…FDA would make a good faith effort to respond within the proposed 90-day timeframe [of the original CFSAN proposed rule], the agency would not be bound…” The timeframe for CFSAN responses is now 180 days, although either is preferable to the 270 day timeframe mentioned by Dr. Benz in July, 2009.
Hopefully, as the GRAS Notification team gets more experience under its belt, pragmatic solutions will be worked out to the satisfaction of everyone.
Also, in the announcement (Section B “Elements of the GRAS Standard”), CVM is following through on previous warnings from the Agency regarding what is meant by GRAS: “…it is the use of the substance, rather than the substance itself, that is eligible for the GRAS exemption…” This statement is consistent with recent statements by the Agency, notably in December of 2009, when CFSAN offered some “Guidance for Industry”. That is, FDA sees the use of an ingredient as being very specific, including concentration in food and for specific food groups (humans) or for a particular named species. A circumstance of use other than what was approved is justifiably regarded as violative and disregards the regulatory process. Further to this issue, FDA has announced it will be taking a bare-knuckled approach, applying the “Responsible Corporate Officer (RCO) doctrine” that allows for civil and criminal prosecution of corporate officers of violative companies; further, for a person to become eligible for prosecution, it is not necessary to prove the person knew they violated the law, only that they violated the law (i.e., ignorance of the law is no excuse). FDA has made it very plain in several venues (See also the Tan Sheet, April 26, 2010 pp. 9 & 11) that manufacturers who do not comply with the regulations do so at their own peril. Given the repetition and frequency of these warnings, we can expect enforcement actions soon.
The announcement also includes a request that the Notifier provide a basis for the determination of GRAS. This is an eminently defensible request. Any successful GRAS determination should be more than a collation of documents. For a successful submission, the petitioner must meet not only the “burden of proof” for the technical element, but the “burden of persuasion” as well. Especially because the cases cited in the announcement all relate to reasons for rejection of a GRAS determination, the need for a persuasive argument of the facts supporting safety-in-use and the presumed consensus of safety, reaches a new level of priority.
As mentioned earlier, the announcement refers to a “pilot program” and is a good start at addressing the long overdue process for addressing GRAS determinations for animal feed ingredients. Our hope is that while the program is here to stay, it is referred to as a “pilot” only because CVM is aware that this is far from a seamless process and considerable fine tuning is needed. However, despite the points raised here, our congratulations to CVM for making this stride forward and our best wishes to Geoff Wong and his team in devising some pragmatic ways for dealing with the inevitable problems that will arise.
 Fed. Reg. 75:31800-31803, 2010
 Fed. Reg. 62:18937-18964, 1997
 Weinberger v. Hynson, Westcott & Dunning, Inc., 412 U.S. 609, 629, 93 S.Ct. 2469, 2483, 37 L.Ed.2d 207 (1973).