TL;DR
- Company: Huons Co., Ltd., a finished-pharmaceutical manufacturer at 100 Bio Valley-ro, Jecheon-si, Chungcheongbuk-do, South Korea.
- Date: Warning Letter 320-26-95 issued June 15, 2026 (MARCS-CMS 724650).
- Classification: CGMP / Finished Pharmaceuticals / Adulterated.
- Key citation: Incomplete and falsified laboratory records (21 CFR 211.194(a)) — a discarded failing bioburden result, manipulated camera timestamps, 1,897 blank uncontrolled lab forms, and logbook pages physically cut out with a knife and replaced with fabricated ones.
- Consequence: FDA rejected the firm's corrective action plan as inadequate after Huons reclassified its own third-party data-integrity audit as an "internal audit" and withheld the report it had originally committed to share.
Every CGMP inspection program is, underneath the paperwork, a bet that a facility's records reflect what actually happened on the floor. In the early 2000s, FDA's own modernization of Part 11 record-keeping rules was built on that same premise: a record is only as good as the process that prevents someone from quietly rewriting it later. Two decades on, the FDA Warning Letter 320-26-95, June 15, 2026, addressed to Site Head Jeong-woo Choo, describes exactly the failure that bet was supposed to prevent: a microbiology team leader who used a knife to cut finished pages out of a logbook and replace them with fabricated substitutes.
| Company | Location | Letter # | Issued | Issuing Office | Inspection Dates | Classification |
|---|---|---|---|---|---|---|
| Huons Co., Ltd. | 100 Bio Valley-ro, Jecheon-si, Chungcheongbuk-do 27159, South Korea | 320-26-95 (MARCS-CMS 724650) | June 15, 2026 | CDER | November 12–21, 2025 | CGMP / Finished Pharmaceuticals / Adulterated |
FDA inspected the Huons facility, FEI 3007163376, from November 12 to 21, 2025. The agency's warning letter, citing 21 CFR 211.194(a), documents a single but layered data-integrity failure in the firm's microbiology laboratory records — one that touched testing worksheets, physical logbooks, and blank uncontrolled forms all at once.
What the inspector found
A failing result, discarded rather than documented. FDA investigators requested testing worksheets from studies the firm had conducted on a compounding tank and its transfer line, covering endotoxin and bioburden testing. Huons initially turned over only part of the data. The next day, the firm's own management told the investigator what had actually happened: the first bioburden run had failed endotoxin limits, and the second run showed significant bioburden growth. A team leader had instructed staff to discard the plates showing that growth and had manipulated camera timestamps to manufacture backdated documentation in their place. The endotoxin failure itself was never documented or investigated at all.
A logbook altered with a knife. During the inspection, investigators found that pages had been removed from a laboratory logbook and replaced with newly fabricated ones made to resemble the originals. Per the warning letter, the microbiology team leader admitted to using a knife to cut the completed pages out, then instructed an analyst to backdate and omit testing information tied to the bioburden samples. This was not a single altered entry; it was a physical rebuild of the record designed to survive a routine review.
Nearly 1,900 forms with no control at all. Separate from the falsified pages, FDA's walkthrough of the microbiology laboratories turned up 1,897 blank, uncontrolled CGMP-related forms. A blank form under document control is traceable — numbered, issued, reconciled. A blank form with no control is an open invitation to generate whatever record a situation later calls for, which is precisely the pattern the logbook and camera-timestamp findings describe.
Why FDA called the response inadequate
Huons responded to its Form FDA 483 on December 15, 2025. The firm's response, as described in the letter, reported that multiple employees involved in the data manipulation are no longer with the company, and that Huons had stood up an internal data integrity team and engaged an independent third-party consultant to audit its operations.
FDA's judgment on that response is direct: it does not provide a CAPA plan with sufficient actions to systemically implement management oversight and assure data integrity. The letter singles out one specific reversal. Huons initially committed to providing FDA the interim and final reports from its third-party data-integrity assessment. The firm later informed FDA that the assessment would instead be classified as an "internal audit," and that the reports would not be provided after all. FDA states plainly that it disagrees with that reclassification.
That sequence is the crux of the letter. Removing the employees who cut logbook pages and manipulated timestamps addresses who did it. It says nothing about how a compounding-tank endotoxin failure went undocumented in the first place, why 1,897 uncontrolled forms existed in a CGMP laboratory, or whether the same conditions that let one team leader falsify records unsupervised still exist for whoever replaced them. Withholding the audit report FDA was told to expect reads, in context, as the firm declining exactly the transparency its own citation was about.
The letter closes with FDA's standard language: failure to promptly and adequately address the violations may result in seizure, injunction, or other legal action, without further notice.
What this means for other manufacturers
A single CFR citation can still describe a systemic failure, and this letter is a clean illustration of that. 211.194(a) covers "complete data derived from all tests," but the violations packed underneath it here span three distinct failure modes: a discarded result, a physically altered record, and thousands of forms with no tracking at all. A compliance team reading a data-integrity citation as a narrow lab-technique problem, rather than a document-control and management-oversight problem, is likely to propose the same kind of personnel-focused fix FDA rejected here.
The audit-reclassification detail also deserves its own line item in any internal playbook. If a firm commits to sharing a third-party assessment with FDA and later decides, after the fact, to withhold it under a different label, that reversal will read to FDA exactly as it reads in this letter: as evidence the firm is still managing what gets disclosed rather than what gets fixed. Any CAPA plan built around removing individuals, without an accompanying audit trail as good as the one FDA is asking to see, is not likely to close a 211.194(a) citation involving deliberate record fabrication.
FAQ
What did FDA cite Huons Co., Ltd. for?
A single CFR citation, 21 CFR 211.194(a), covering incomplete laboratory records. Within that citation, FDA documented a discarded failing bioburden result with manipulated camera timestamps, an undocumented endotoxin failure, 1,897 blank uncontrolled lab forms, and logbook pages physically cut out and replaced with fabricated substitutes.
Did someone at the company admit to falsifying records?
Yes. Per the warning letter, the firm's microbiology team leader admitted to using a knife to remove completed logbook pages and instructed an analyst to backdate and omit bioburden testing information on the replacement pages.
Did FDA accept the firm's corrective action plan?
No. FDA found the response inadequate, noting it lacked a CAPA plan sufficient to implement systemic management oversight of data integrity. FDA also stated it disagreed with the firm's decision to reclassify its third-party data-integrity assessment as an "internal audit" and withhold the report it had originally committed to provide.
What products does this warning letter cover?
The letter addresses CGMP violations at the firm's finished-pharmaceutical manufacturing facility in Jecheon, South Korea, tied to laboratory testing on a compounding tank and its transfer line. The public letter does not name specific finished-drug products beyond describing the facility's CGMP scope.
Source: FDA Warning Letter 320-26-95, June 15, 2026. Byline: Andy Gaber. Published 2026-07-08.