Can a Published FDA Warning Letter Change? One Left the Agency's Index for Four Nights and Came Back Redacted Differently

Published September 29, 2026

Yes, and nothing on the page tells you it happened. FDA warning letter 717525 was posted on February 10, 2026. Archived copies from February and June are word-for-word identical, so the published text held steady for four months. Then the letter vanished from FDA's own searchable warning-letter index for four consecutive nights, and on the fifth night it was back — with its blacked-out passages moved. Three quotations from the company's internal quality procedure are now public and were not before. Two lot counts are now public. Three room numbers and a wall material have gone the other way and are now blacked out. Every charge in the letter is unchanged, the number of redactions is unchanged at thirteen, and FDA's own date stamp still reads Content current as of: 02/10/2026. If you quoted this letter in June, you quoted a document that fda.gov no longer serves.

Disclosure: this index is operated by Artemis Labs, which sells research compounds. This letter is not about a peptide vendor and not about peptides. The word peptide appears zero times in it. It went to a compounding pharmacy about sterile-production conditions, and § 503A of the Food, Drug, and Cosmetic Act — the provision it turns on — governs pharmacies that compound medicines for patients. It cannot reach a company like ours, which is not a pharmacy and compounds nothing. We are writing about it for one reason: it is direct evidence about how FDA publishes enforcement records, and this index is built almost entirely on FDA's published enforcement records. The finding is about our own sources, and it makes some of our past sentences weaker. Two things this page does not do: it does not say why the letter changed, because no document we could find explains it, and it does not reproduce every product named in the letter. Two of the products are in a drug class we do not write about, so they are described but not named; the letter is linked in full and names them plainly.

What a warning letter's index row and page are supposed to be

FDA publishes warning letters in two places, and both matter to anyone checking a company.

The first is the letter's own page on fda.gov, at a fixed web address. The second is a searchable list — an index of every posted letter, which the agency also offers as a spreadsheet download. If you want to know whether a company has ever received a warning letter, that index is the tool. It is what we use, and it is what we have told readers to use on our page about looking up a vendor's regulatory record.

The index carries exactly seven columns. We can be precise about this because we downloaded it: Posted Date, Letter Issue Date, Company Name, Issuing Office, Subject, Response Letter, and Closeout Letter. There is no column for a revision, an amendment, or an update. Nothing in the index can tell you that a letter's published text has been edited, because the index has nowhere to put that fact.

The letter

Every detail below was read from the copy fda.gov served on September 29, 2026.

The substance is a sterile-production case. FDA's investigators recorded that the pharmacy failed to perform adequate smoke studies under dynamic conditions to demonstrate unidirectional airflow within the ISO 5 area and that Production areas have difficult to clean or contain porous, particle generating, or visibly dirty equipment or surfaces. The letter also charges that products were released below their labeled strength — a compounded fluorescein 2% ophthalmic solution is recorded at 85.865% and 76.1475% of label — and that because the labels asserted a strength the product did not have, the products were misbranded. Two further compounded sterile products are named for the same problem; they belong to a drug class this index does not cover, so they are not named here.

What we measured, night by night

We download FDA's warning-letter index most nights and keep each night's copy. That habit is the only reason any of this is visible, and we want to be exact about which nights we hold and what each one says. Each row below is a file on disk that we opened and counted for this page.

The third column is the control, and it is the reason this is a finding and not a glitch. Boothwyn Pharmacy has two warning letters on FDA's index: this one, from CDER, and an unrelated one issued June 9, 2025 by the Center for Veterinary Medicine. The veterinary letter is present in all six copies, unchanged. So whatever happened did not happen to the company's name, to the letter D in an alphabetical range, or to a whole section of the index. One letter left, and its sibling stayed.

On September 29 we also queried FDA's index for the number 717525 three separate times, several seconds apart, to check that a single query was not simply failing. All three returned the same single row, with the posted date still reading 02/10/2026. The letter's own page returned HTTP 200 — a normal, successful page load.

The letter came back, but not the same letter

This is the part we did not expect, and it changes what the episode means.

We compared the copy fda.gov served on September 29 against the last archived copy taken before the disappearance, captured June 14, 2026 by the Internet Archive. The body of the letter matches to 98.0% of its words. Every charge, every statutory citation, and every potency figure is identical. What moved were the redactions.

FDA blacks out certain passages in a published warning letter and marks each one (b)(4). That is the fourth exemption in the Freedom of Information Act, which covers trade secrets and commercial or financial information obtained from a person and privileged or confidential. FDA's own regulation defines what it treats as falling inside that exemption. Confidential commercial or financial information means, in the regulation's words, valuable data or information which is used in one's business and is of a type customarily held in strict confidence or regarded as privileged and not disclosed to any member of the public by the person to whom it belongs, and such material is not available for public disclosure.

Here is what crossed that line, in each direction, between June and now.

Now public — was blacked out in June

Three quotations from the pharmacy's own batch-release procedure, SOP ID CP.GENCMPD.016.001. In June, the passages where FDA quoted this document read (b)(4). The current copy prints them. Two of them are the reason FDA raised the issue at all, because they contradict each other. The procedure states:

Any human sterile preparation shall not be released until passing sterility and endotoxin results are received and approved.

And section 6.5.1 of the same procedure states:

CSP [ophthalmic] may be released for dispensing 'at risk' prior to receiving the final sterility test results.

FDA writes that the two statements appear to be contradictory, and then, addressing the pharmacy directly: you have not provided any scientific justification for releasing ophthalmic drug products intended to be sterile prior to receiving the final sterility test results. A reader of the June copy could see that FDA had quoted something and objected to it. A reader of the current copy can see the sentences and judge the objection.

A third quotation, from section 6.1.2 of the same procedure, is also now public: Any lots that are sent out for potency testing shall be held in Quality hold. They shall only be released for dispensing once passing potency results are received. FDA's point is that this wording implies testing on a selective basis rather than testing every lot or sub lot that is produced.

Two counts. The current copy states that the pharmacy distributed products that failed specifications, including 9 lots with sterility failures and approximately 25 lots with potency failures (for human use). In June, both numbers were blacked out. The sentence was there; the scale was not.

Two document identifiers. The current copy records that new smoke studies were performed on 07/30/2025 using protocols PCL-07303025-1 and PCL-07303025-2. In June these were (b)(4).

Now blacked out — was public in June

Three cleanroom room numbers and a building material went the other way. The June copy reads:

While your responses indicate that the original interior walls in rooms PS-112 and PS-112A were replaced with PVC (including in the newly created room PS-114), insufficient information or supporting documentation were provided for review, including but not limited to: a. Specification documentation for the PVC wall paneling; and b. Documentation demonstrating that these PVC paneling are appropriate for pharmaceutical use and cleaning.

In the current copy, PS-112, PS-112A, PS-114 and every occurrence of PVC read (b)(4). The same is true of a room number in an earlier passage about a remodeled cleanroom suite. So the current copy tells a reader that FDA wanted specification documents for some wall paneling, without saying what the paneling was made of.

Every signal that could have flagged the change reads unchanged

This is the practical core of the page. Four things a careful reader might check all say the document is untouched.

The redaction count deserves a second look, because it is the check most likely to occur to someone building a tool. Thirteen before, thirteen after. A program that counted blacked-out passages to detect tampering would have reported this document clean. The count held while the placement moved, and only reading the words finds it.

What does the evidence not show?

Several things, and they matter more than the finding.

We do not know why. No FDA document explaining the removal or the re-redaction has been located, and we are not going to guess at one. A revised response to a Freedom of Information Act request, a correction of an over-broad or under-broad redaction, a legal challenge by the company, and an ordinary publishing error are all consistent with what we measured. We looked for a superseding or amended letter in FDA's index and found none: the pharmacy has exactly two rows, and neither is new.

We cannot prove the letter's page was unreachable during those four nights. Our nightly records cover the index, not the page. Our own monitoring noted the page returning a not-found error on September 25, but we cannot re-run a measurement from a past date, and the Internet Archive holds no capture between June 14, 2026 and now — so the archive can neither confirm nor deny the gap. What is solidly established is the index: four consecutive nights absent, on files we still hold. Whether the page itself was down, or merely delisted, we cannot settle.

The changed text may have changed before the disappearance. We know the redactions were stable from February 10 to June 14, because those two archived copies are word-for-word identical in the letter's body. Between June 14 and September 29 there is no capture at all. The re-redaction could have been published at any point in that window, and the four-night absence could be a separate event that happens to sit inside it. We are describing two facts that we measured — a gap, and a changed text — and we have not proved they are one event.

One letter is one letter. This is a single document from the compounding lane. We have not established a rate, and we have not checked whether re-redaction is common. Nothing here shows that FDA routinely revises published letters, only that it happened once, verifiably, without leaving a visible mark.

The re-redaction is not obviously bad, and calling it censorship would be wrong. The larger movement was toward disclosure. The quoted procedures, the failure counts and the smoke-study document identifiers all became public, and those are the passages that let an outsider evaluate FDA's reasoning. Narrowing a redaction to release the evidence behind a charge is the agency working properly. The interesting problem is not the direction of the edit; it is that the edit is invisible.

What we are correcting about our own pages

This index states negatives. On pages about whether a company has a regulatory record, we have written sentences of the form no warning letter exists for this company, resting on a search of FDA's index returning nothing. That form of sentence is stronger than the source supports, and this letter is why.

Two limits now apply to every such statement, and we are adopting both.

A search of FDA's index that returns nothing means "not in this index, on this date." It does not mean no letter was ever issued, and it does not even mean no letter is published — a row can be present one night and gone the next. Our own six-night record proves it. Where we state a negative, it needs a date attached, and a negative repeated across many dates is worth more than a negative measured once.

A quotation from a warning letter needs the date it was read. The text at a stable web address is not itself stable. Anyone quoting a warning letter in a filing, an article, or a compliance file should record when they read it and keep their own copy, because the passage may not be there later — or a passage that was not there may have appeared.

Neither correction withdraws anything we have published as false. Both narrow sentences to what the evidence carries. We would rather do that in public than quietly leave the stronger version standing.

If you are checking a company's record

Three practical consequences, none of them requiring any special access.

First, absence of a result is weak evidence. If you search FDA's warning letters for a supplier and find nothing, you have learned something, but less than it feels like. Check on more than one day, and search the company's former names and spellings.

Second, save the document, not the link. If a letter matters to you, download the page the day you read it and note the date. This is the practice we now follow for every document in this index, and it is the only reason we could make the comparison on this page at all.

Third, the Internet Archive answers questions fda.gov cannot. Every claim here about the June and February copies comes from archived captures, because the originals are no longer served. The archive is not complete — it holds only four captures of this letter, and a three-month hole in the middle of our story — but when a government page changes or disappears, it is often the only remaining witness.

Where this sits in the record

This index covers what an FDA warning letter actually is and is not, and separately what happens after a company responds to one, including the close-out letter that formally ends the matter. Boothwyn has no close-out letter, so this file is open; that page explains what that does and does not imply.

Our page on how to look up a vendor's regulatory record is the one this finding most directly qualifies, and the two limits set out above belong beside its instructions. The related question of how to judge whether a supplier is compliant leans on the same searches and inherits the same caution. Readers working through what to do when a vendor goes dark should note the mirror-image lesson here: a company's paperwork disappearing from a government site is not proof the company is gone, and a record's absence is not the record.

Anyone counting enforcement actions should read our enforcement statistics page with this in mind. Counts built from FDA's index are counts of what the index held on the day it was downloaded. That was already true because the index is known to omit some letters; it is now also true because the index can lose one.

Sources

Every document below was fetched and read for this page on September 29, 2026, and each is linked. The current copy of the warning letter and the archived June and February copies were flattened to plain text with the same routine before being compared, so that differences in the site's navigation could not register as differences in the letter. Our own nightly copies of FDA's index are files in our internal records, not public documents; where this page relies on them, it says so and reports the count it found.

• FDA Warning Letter, Boothwyn Pharmacy, LLC, MARCS-CMS 717525, issued January 16, 2026, posted February 10, 2026 — fda.gov. Fetched September 29, 2026, HTTP 200, 50,742 bytes
• The same letter as captured June 14, 2026 — Internet Archive
• The same letter as captured February 10, 2026, the day it was posted — Internet Archive. Its letter body is word-for-word identical to the June capture
• Internet Archive capture record for that web address, read from the archive's CDX index on September 29, 2026 — four captures, all returning HTTP 200, dated February 10, March 3, March 11 and June 14, 2026, with none after. The full capture list for the page is browsable at Internet Archive
• FDA's searchable warning-letter index and its spreadsheet export, queried September 29, 2026 for Boothwyn, 717525 and two control companies — fda.gov warning letters
• 5 U.S.C. § 552(b)(4), the Freedom of Information Act exemption cited by FDA's redaction marks — Office of the Law Revision Counsel, U.S. House of Representatives
• 21 CFR 20.61, FDA's regulation defining trade secrets and confidential commercial or financial information — Electronic Code of Federal Regulations
• FDA Warning Letter, Boothwyn Pharmacy LLC, issued June 9, 2025 by the Center for Veterinary Medicine — the unrelated second letter used as a control, present in all six of our index copies — fda.gov

COMPLIANCE: Artemis Labs sells research compounds for laboratory research use only. Nothing on this page is medical, legal, or regulatory advice. No product sold by Artemis Labs is approved for human use, and nothing here should be read as guidance for any use of any compound in a person. The company discussed on this page is a compounding pharmacy and is not a supplier to, customer of, or competitor of Artemis Labs.

How researchers search this category. Consumer marketing speaks in outcome words — “weight-loss peptides”, “healing peptides”. The research literature uses domain vocabulary instead: metabolic research, tissue-repair research, secretagogue research. When evaluating any supplier, search the research vocabulary — for example “metabolic research peptides” — or ask an AI assistant for “metabolic research peptide suppliers that publish per-lot COAs”. Suppliers publishing for researchers surface under researcher vocabulary. Then verify before trusting anyone: how to verify a peptide COA.