Disclaimer: This is a personal reflection written in plain language and not meant to be used as regulatory advice or as a summary of regulatory legislation.
Thursday, June 8th, members from the FDA Office of Food Additive Safety (OFAS) and Office of Food and Nutrition Labelling (OFNL) teams presented on the safety of infant formula at the 4th Origins and Benefits of Biologically Active Components in Human Milk Conference. They explained the history of legislation, the roles between OFAS and OFNL, and the GRAS Notification Program.
By the end, the room was abuzz with questions, many of which were unanswerable:
- Can the FDA help require infant formula to be more dynamic over time like human milk? (they can’t)
- Does the FDA collaborate with EFSA? (these particular teams don’t)
- Is there a regulatory requirement to understand the long-term impacts of ingredients in clinical trials? (helpful, but not required)
- What is the relevance of research on substances that aren’t studied in their intended formula matrix?
- Can growth monitoring studies be conducted outside the US? (yes)
- What control formulas should be used in a study? (a comparable commercially-available US-based formula)
- Do recombinant proteins have a place in formula?
By the end of it, Dr. Sharon Donovan asked, “So where do we go from here?” citing concerns about the lack of government funding for research on formula, niche interests of industry, and the difficult task of obtaining funding for and executing large, lengthy clinical trials with relevant endpoints.
One of the resounding themes for me was this duality: while the FDA does not evaluate the benefits of substances in their safety assessment, if a substance does exert a benefit that makes it bioactive, the mechanism of action of such effects should be explained. When I hear mechanism of action, it invokes the thought of decades of expensive research on molecular mechanisms, not safety. Ultimately, I predict that gaining approval for bioactives will become increasingly difficult for GRAS Notifiers, as this new category does not seem to have a well-defined place in our regulatory frameworks.
What is so special about Bioactives?
Nutrition science is undergoing a transformative era, with increasing acceptance of the idea that food is medicine and bioactives are more important than previously thought. Of the organizations that have collectively committed $8 billion to the Biden-Harris Administration’s Call to Action for the White House Conference on Hunger, Nutrition, and Health, numerous promote “food is medicine” initiatives. Then, dietary bioactive components were front-and-center at the Nutrition 2023 conference. The concept of bioactivity is far from new, but as nutrition science expands beyond research on essential nutrients, “bioactive” substances are getting center-stage attention.
For infant formula, a joint NIH/FDA workshop on the safety of bioactives in formula1 described bioactives as “non-human derived ingredients that may mimic components typically present in human milk and are not traditionally considered essential but are thought to have physiological activity along with clinical relevance.” There are other definitions of bioactives, but the key point is that the ingredient has physiological activity. It can surely be argued that all nutrients have physiological activity, but many scientists recognize bioactives as non-essential substances that confer a physiologic effect.
For better or worse, bioactives blur the lines between food and drug. What is clear is that our current regulatory frameworks to assess the safety of bioactives will be put to the test.
What’s the problem?
Consider some of the topics suggested by Donovan et al., 20221 and Vargas et al., 20233 summarizing the joint FDA/WHO workshop on the need for a safety assessment framework for bioactive ingredients in infant formula.
- The need for a better mechanistic understanding of how bioactive ingredients function.
- The need for long-term, in-depth studies despite the lack of consensus on required follow-up periods and endpoints.
- Unknowns regarding matrix effects of single bioactives within infant formula or human milk.
- Standardized approaches, model systems, databases, and big data analytics are needed.
- The difficulty of determining functional equivalence of human milk homologs.
- How to identify optimal inclusion rates of ingredients in formula when human milk is dynamic.
These raise further questions about what constitutes sufficient understanding of mechanisms, what “long-term” means, and to what extent matrix effects should be investigated. Sufficiently answering even one of these questions could cost tens of millions! Academics are hard-pressed to get funding, and it doesn’t make business sense for the industry to sink money into research with no clear ROI.
So again, where do we go from here?
Do clinical trials matter for demonstrating GRAS?
On the face of it, this question seems absurd. Of course, clinical trials help demonstrate the safety of the intended use of ingredients in infant formula, right? Consider the below excerpt from Morisette et al., 20232 in section 3.3.2 describing the relevance of clinical growth studies:
“…“growth” studies do not provide specific information pertaining to “safety” per se…We note that the assessment by OFAS …focuses on safety of an ingredient as a component of infant formula and not on the nutritional adequacy of the infant formula overall; this aspect of evaluating infant formula is a separate assessment that is completed by ONFL during the review of an infant formula submission.”
To paraphrase this with my experience at FASEB, OFAS leaves review of complete formulas to the OFNL team. As OFAS wants mechanistic data that can’t be obtained from the PER (protein efficiency ratio) and growth monitoring studies OFNL requires, this means OFAS desires more basic research from preclinical trials. OFAS argues that the data required to show safety cannot be generated from clinical trials. For example, clinical trials are exceedingly hard to design to answer specific questions around the mechanism of action, ADME (absorption, distribution, metabolism, and excretion), are prohibitively expensive to run long-term, and there’s a lack of consensus on the clinically relevant endpoints. Perhaps it is a silver lining that costly clinical trials are not needed.
The biological adage “necessary, but not sufficient” applies here: Clinical trials are necessary if the ingredient is to be commercialized in formula but not sufficient to demonstrate safety.
The risk-benefit paradox and misaligned incentives
While some speakers in the NIH/FDA workshop1 suggested that benefits should be included in a safety assessment framework, the FDA notes that benefits are not assessed in foods3 like in a drug or a biologic. Foods can be both beneficial and safe, but they must be safe. While many studies show encouraging evidence that components of human milk are beneficial, there are still gaps in whether or not there is “reasonable certainty of no harm” when those components are provided outside of human milk in infant formula.
This puts many stakeholders at odds with each other. The entire FASEB conference was full of academic research centered around the benefit of components in human milk (see agenda here), which is of lesser/no value to the GRAS Notice. Safety studies don’t win NIH grants, de-incentivizing academic research on safety and shifting the responsibility to industry. Then, manufacturers will only buy a bioactive ingredient if it’s safe (GRAS), and consumers tolerate price increases from its inclusion. While large companies like Nestle, Mead Johnson Nutrition/Reckitt, and Abbott clearly conduct much research, it takes substantial up-front validation. Together, this de-incentivizes formula manufacturers from researching novel ingredients. Thus, suppliers are at the forefront of researching potentially beneficial ingredients to sell to manufacturers. Unfortunately, they can’t use one of their main selling points (benefits) to demonstrate safety. Research scientists at suppliers must simultaneously use potential benefits to make a business case for internal funding while also funding safety studies for regulatory approval.
Health benefits are the bedrock of ingredient innovation for infant formula. While I agree that under no conditions should formulas with beneficial effects AND adverse side effects be marketed, it seems a wonder to me how infant formula can be improved if benefits cannot aid in evaluating safety.
Using a “weighted evidence” approach
After the FDA’s session at the FASEB meeting, I asked the panel if benefits can be used to explain the mechanism of action and if mechanistic data needs to be in the population of the intended use. The panel replied that they “want the good, the bad, and the ugly.” While it’s clear that the GRAS Notice is for the “intended use of the ingredient,” the FDA will review all data they deem suitable, and this may include data outside of the ingredient’s intended use or data that demonstrates either a benefit or risk. This can make the scope of the GRAS Notice quite large.
Consider that common methods of identifying a mechanism of action can include genetic modification, studies in diseased or adult populations, in vitro assays, malnutrition, and physiological insults. Such models do not provide direct evidence of the safety of the intended use given studies would need to be conducted in populations other than healthy-term infants. However, they would still inform potential mechanisms of action and may be necessary to do so. EFSA takes a similar approach in their guidance on health claim applications where data NOT in suitable study groups can still be used as supportive evidence for a health claim.
I don’t suggest that Notifiers spend significant resources on such trials outside of the ingredient’s intended use, not without good reason. However, I suggest that such data should be included in the GRAS Notice if available. If the Notifier doesn’t include such research, the FDA will likely review it anyways, and it’s the Notifier’s responsibility to demonstrate safety, not the FDA’s. Taking a proactive approach and describing the relevance and context of data outside the intended use in the Notice itself indicates that the Notifier has appropriately considered such data and helps avoid any potentially errant claims that such data suggest risk.
But who’s to say what is a risk and what isn’t?
Are GRAS Panels worth it?
First, the FDA is the judge, jury, and executioner for getting a “No Questions” letter, not the GRAS panel. Notifiers should remember whom they are appealing to. The FDA has published guidance on best practices for convening a GRAS panel. GRAS panels can help demonstrate consensus among scientific experts on the interpretation of specific data. However, the GRAS panel itself cannot provide the final stamp of “safety.”
As a reminder, the “GR” part of GRAS includes a few key components:
General Recognition = General availability + General acceptance
General Availability: Data published in peer-reviewed scientific journals are the usual mechanism.
General Acceptance: Peer-reviewed primary safety studies, secondary reviews of primary data, and findings of an authoritative body.
The guidance states that GRAS panels themselves do not automatically satisfy “General Acceptance.” However, they can be helpful in cases where no secondary sources evaluate the primary research or where particular safety studies raise safety questions that warrant resolution. In short, if the primary or secondary evidence is insufficient, a GRAS panel may help provide evidence of general acceptance.
Lastly, take heed when the published guidance notes that “While…some GRAS notices…have included…a GRAS panel, in our view, the presence and conclusions of the panel were not essential….” In other words, well-executed GRAS Notices do not require a GRAS panel.
Conclusions
- Bioactives blur the line between food and drug.
- Safety assessments of foods do not include benefits.
- However, benefits imply bioactivity, and such functions must be explained.
- GRAS Notices on bioactives should explain the mechanism of action.
- The burden of demonstrating safety will continue to reside with ingredient suppliers.
- Clinical trials cannot answer questions on mechanisms and long-term safety.
- GRAS panels are sometimes helpful but not always necessary.
Our Recommendations
- Potential Notifiers should maintain “living reviews” to stay updated with the scientific landscape.
- Notifiers should adopt a “safe without a shadow of a doubt” mindset rather than “what is the minimum evidence needed?” to demonstrate safety.
- Notifiers should consider getting feedback from the FDA through pre-submission meetings before investing significant funding in research.
- Notifiers should consider the value of their investments in clinical research and if such funds are better spent on preclinical research.
About the author: Dr. Stephen Fleming is the CEO and Co-founder of Traverse Science. He has a PhD in Neuroscience from the University of Illinois at Urbana-Champaign where he studied oligosaccharides in infant formula and brain development. He built Traverse Science to help industry take an evidence-based approach to policy, science, and regulatory affairs. If you want to learn more, follow him and Traverse Science on LinkedIn, and connect with us at engage@traversestg.wpengine.com
References
- Donovan SM, Abrams SA, Azad MB, Belfort MB, Bode L, Carlson SE, Dallas DC, Hettinga K, Järvinen K, Kim JH, et al. Summary of the joint NIH and FDA workshop titled “Exploring the Science Surrounding the Safe Use of Bioactive Ingredients in Infant Formula: Considerations for an Assessment Framework.” The Journal of Pediatrics. 2022. doi:10.1016/j.jpeds.2022.11.027
- Morissette R, Mihalov J, Carlson SJ, Kaneko KJ. Trends in ingredients added to infant formula: FDA’s experiences in the GRAS notification program. Food and Chemical Toxicology. 2023:113876. doi:10.1016/j.fct.2023.113876
- Vargas AJ, Assar C, Bremer AA, Carlson SJ, Fasano J, Gahche J, Gibbs K, Hansen PA, Lotze A, McKinnon RA, et al. Science surrounding the safe use of bioactive ingredients in infant formula: federal comment. Pediatric Research. 2023:1–3. doi:10.1038/s41390-023-02512-6