When we wrote about The Protein Brewery’s two regulatory outcomes on 13 August, we ended with an admission: the FDA’s response letter could not be retrieved, the quoted deficiencies came from AgFunderNews reporting rather than our own reading, and we had not independently verified the wording.
We have now retrieved and read the letter. It is two pages, signed on 3 March 2026 by Susan J. Carlson, PhD, Director of the Division of Food Ingredients. The quoted wording is accurate. But the letter contains four dates that were not in the reporting, and the intervals between them change what the episode means.
The dates, and what they measure
| Date | Event | Day of clock |
|---|---|---|
| 14 March 2025 | FDA receives the notice | Before filing |
| 19 August 2025 | FDA files the notice | Day 0 |
| 11 February 2026 | FDA emails about the additional information needed; lists five deficiencies | Day 176 |
| 13 February 2026 | FDA receives the request to cease evaluation | Day 178 |
| 15 February 2026 | Statutory 180-day response deadline | Day 180 |
| 3 March 2026 | FDA grants the request; notice closed | After withdrawal |
Under 21 CFR 170.265(b)(1), FDA “will respond to you by letter” within 180 days of filing, extendable by 90 days “on an as needed basis.” The clock starts at filing, not receipt.
Filing was 19 August 2025. Day 180 is therefore 15 February 2026.
FDA raised its deficiencies on 11 February 2026 — day 176. The Protein Brewery’s request that FDA cease evaluating the notice arrived on 13 February 2026 — day 178.
Two days.
Practical consequence: the interval that matters in a GRAS notice is not the one companies report. Everyone tracks filing-to-response. Almost nobody tracks the two other intervals that decide the outcome — receipt-to-filing, and deficiency-to-decision. Both are in this letter, and both were short in the direction that hurt.
Five months before the clock even started
FDA received the notice on 14 March 2025 and filed it on 19 August 2025. That is 158 days in which the submission sat in initial evaluation, before the 180-day review period began.
Section 170.265(a)(1) describes this as an initial evaluation “to determine whether to file it as a GRAS notice.” It has no stated time limit. So the real elapsed time from submission to closure was 14 March 2025 to 3 March 2026 — 354 days — of which the statutory review window accounted for less than half.
For anyone modelling a US route to market against an EU one, that is the number to use. Not 180 days. Not 270 with the extension. The observed figure here is 354 days to reach no decision at all.
What the notice actually asked for
The deficiency that reads most consequentially is the second: “insufficient data to support safety at the proposed dietary exposure levels.” The inventory entry makes clear what those levels were.
The notice sought use of Rhizomucor pusillus mycelial biomass “as a source of protein and fiber at levels up to 100%” across twenty-three comma-separated food categories: cakes, breads, rolls, muffins, tortillas, waffles, bagels, brownies, crackers, breadsticks, non-milk based protein and nutrition powders, hot breakfast cereals, non-dairy yogurts and non-dairy cheese, cereal bars, protein bars, nutritional bars, pasta, noodles, nut spreads, chips, cereal and root-based snacks, chocolate confectionary, and chocolate and other sweet spreads. Excluded: infant formula, foods intended for infants, and products under USDA jurisdiction.
A notice claiming up to 100% inclusion across that range implies a dietary exposure ceiling close to the whole of a person’s intake from those categories. The agency’s objection to the exposure data is not a technicality about how a table was laid out. It is an objection proportionate to what was asked for.
Practical consequence: the breadth of an intended-use statement is a regulatory risk variable, not a marketing one. A narrower first notice — fewer categories, a stated maximum inclusion rate — asks the agency to be satisfied about less, and is the standard way a second attempt differs from a first.
The five deficiencies, verbatim
From the letter, describing the 11 February 2026 email:
significant design flaws in animal studies that compromise data interpretation; insufficient data to support safety at the proposed dietary exposure levels; inadequate safety assessment of key compositional elements with known antinutrient properties; inadequate bioinformatics and in silico approaches; and absence of organism-specific secondary metabolite characterization.
The letter then states: “Given the extent of the information needed to continue our evaluation, we recommended that The Protein Brewery request that we cease our evaluation of the notice.”
That sentence is the agency’s own characterisation of severity. FDA did not run out of time. It concluded the gap was wide enough that stopping was the appropriate course, and said so.
The company’s account, set against the letter
CEO Thijs Bosch told AgFunderNews in June that the deficiencies were “straightforward to address,” required no studies to be redone, and were a matter of presenting existing data in line with the EFSA submission. He offered an illustration: in the original filing “there was a typo in the name of the type of rats used in a tox study.” He said the company stands behind “our process, our strain and our studies,” expects a no-further-questions letter in Q1 2027, and expects first US products in late 2026.
Both accounts can be partly true. A misnamed rat strain is exactly the kind of error that would surface under the heading “significant design flaws in animal studies that compromise data interpretation” if a reviewer could not confirm which animal model was used — and it is also exactly the kind of error that is genuinely fixed by correcting a document rather than running a new study.
But the letter names five deficiencies across five distinct disciplines: toxicology study design, exposure assessment, compositional and antinutrient safety, bioinformatics and in silico methods, and secondary metabolite characterisation. A typographical explanation covers the first at most. “Absence of organism-specific secondary metabolite characterization” describes something missing, not something mislabelled.
And the two-day interval is hard to reconcile with a routine correction. A company that believed the objections were presentational had, on the face of 170.265(a)(4), the option of submitting a timely amendment — FDA “will consider any timely amendment,” though only “if we deem that doing so is feasible within the timeframes.” With four days left, it plainly was not feasible. The alternative to withdrawing was a substantive FDA letter on the public record within two days. Withdrawal produced a letter that says the agency ceased evaluation at the notifier’s request.
We are not saying the company was wrong to withdraw. On these dates it is the obvious commercial decision, and the regulation exists to permit it. We are saying that “temporary setback” and “the alternative was an adverse letter in 48 hours” describe the same event, and only one of them is in the public account.
The reminders FDA chose to include
The letter’s closing paragraphs are boilerplate, but the selection is not neutral. FDA reminds the company:
- of “a manufacturer’s responsibility to ensure the safety and regulatory status of the substances that it markets for use in food”
- that use of a non-GRAS substance “must have pre-market approval by FDA” under 21 CFR 170.30(g)
- that under the signed certification in 21 CFR 170.225, the company “agrees to make all data and information regarding its GRAS conclusion available to FDA upon request”
The Protein Brewery self-affirmed GRAS in March 2024 and, per Bosch, is updating its self-affirmation documentation so customers can use the ingredient while it waits. Selling on self-affirmed GRAS is lawful and common — we counted the scale of the practice when FDA proposed to end it in its mandatory notification rule.
What is unusual here is the sequence. The agency has stated on the record that it had insufficient data to support safety at the proposed exposure levels, and the ingredient is expected on the US market this year under a determination the agency has not reviewed. Both facts are public. Any US customer’s regulatory affairs function will find both.
Practical consequence: if you are a formulator considering this ingredient in the US before Q1 2027, the diligence question is no longer “is it GRAS.” It is whether your own risk position is comfortable with a self-affirmed conclusion covering exposure levels the agency has specifically said were not supported, and whether your supply agreement allocates that risk.
The counter-argument
The serious case for the company’s reading is EFSA.
The same organism, the same strain and substantially the same data package cleared a full EFSA safety assessment, and on 17 June 2026 the European Commission adopted the implementing regulation adding Rhizomucor pusillus mycelium to the Union list. EFSA is not a soft regulator, and its 2015/2283 novel food assessment covers production process, composition, stability, specifications and toxicology.
If a package satisfies EFSA and fails FDA at the first pass, the most economical explanation is not that the science is unsafe. It is that the two agencies want different dossiers — different study designs, different exposure modelling conventions, different expectations on bioinformatic allergenicity screening and secondary metabolite work. Bosch’s stated plan is precisely to align the FDA submission “in line with all of the EFSA related data,” which is a coherent response to that diagnosis.
The €18 million Series B extension, led by ABN AMRO Sustainable Impact Fund with Invest-NL, Novo Holdings, Madeli and BOM, was raised after the withdrawal. Investors with access to the data room priced it as survivable.
The counter to the counter is narrow but real: EFSA clearance does not retroactively supply “organism-specific secondary metabolite characterization” if that work was not done. If it was done and was simply presented differently, the resubmission should be quick. Q1 2027 is the company’s own test of that, and it is nine months from the interview.
What we could not establish
- The 11 February 2026 email itself. The letter describes and quotes it. We have not seen the email, and it does not appear in the public inventory.
- Whether FDA offered an extension. Section 170.265(b)(1) permits a 90-day extension, and (b)(2) requires FDA to notify the notifier in writing “no later than within 180 days of filing.” The letter does not mention an extension being offered, granted or declined. Its silence is not evidence either way.
- Whether the company considered a timely amendment. Inferred as impractical from the four-day margin, not stated by either party.
- The full GRN 1268 notice. The releasable version runs to 3.3 MB on the regulations.gov docket. We read the inventory entry and the response letter in full; we did not read the underlying dossier, so we cannot say from our own reading whether the secondary metabolite work exists in it.
- Whether a pre-submission meeting has taken place. Bosch said in June one was planned “in the coming weeks.” We found no public record of it.
- US launch status. Bosch expected first US products in late 2026. As at 4 September 2026 we found no announced US product containing Fermotein.
What to watch
- A new GRN number for the same organism. Under 21 CFR 170.260(b), cited in the letter, the company may file again. A fresh notice will appear in the inventory with a new number, and its intended-use statement will be the tell: if the twenty-three categories or the “up to 100%” ceiling narrow, the exposure objection was the binding one.
- The filing date on that notice, not the submission date. On this precedent the gap was 158 days. Whether that repeats tells you whether the first initial evaluation was slow for reasons specific to this dossier.
- Q1 2027. Bosch’s stated expectation of a no-further-questions letter. It is a dated, testable claim and we will check it.
- Whether any named US brand launches on the self-affirmed conclusion. That is the point at which the regulatory question becomes a commercial one for somebody other than the notifier.
One correction to our own record, made explicitly: in August we wrote that the FDA closed GRN 1268 “after telling it the notice had deficiencies.” That is right, but it understates the compression. The interval was two days, inside a window with four days left. We have the document now, and the document is more specific than the reporting was.