“Media fills cannot justify unsuitable manufacturing lines” — the response that fixed the test, not the line
483 response filed 11 March 2026 · Citations 21 CFR 211.192, 211.113(b), 211.110(a), 211.42(c)(10)
A three-week inspection of Fresenius Kabi’s sterile injectables site in Melrose Park, Illinois found four media fill failures on two aseptic lines across 2024 and 2025, and batches of famotidine injection released on passing sterility and endotoxin results while patients were reporting apparent pyrogenic reactions. The firm filed a detailed response on 11 March, and much of it was sensible: better simulations, better tracking, outside help.
FDA still called it inadequate on every count, and the reason is instructive. Almost every commitment improved how the lines are measured. FDA’s concern was the lines themselves. A response that upgrades the test has not changed the thing being tested.
What FDA found
On aseptic process simulation, FDA said the firm’s “media fills failed to accurately simulate aseptic manufacturing operations.” They “did not adequately represent” maximum aseptic intervention counts, inherent interventions such as stopper additions and vial loading were not documented in batch records, and “the number of units in your media fills was insufficient.”
Behind that sat four failed media fills: two in September and October 2024 on a line with limited barrier protection, and two in April and October 2025 on a restricted access barrier system (RABS) line. The firm’s investigations attributed each to “deficient aseptic techniques” by operators and did not ask why the aseptic observer programme had not caught those techniques first.
On famotidine injection, the firm received adverse event reports in 2025 and 2026 including apparent pyrogenic reactions. Manufacturing had multiple bioburden action-limit failures, with gram-negative confluent growth, yet batches were released on passing sterility and endotoxin results. Retained product tested during the adverse event investigation failed endotoxin, and three batches were recalled.
- 2024–2025
- Four media fill failures on two aseptic lines.
- 27 Jan – 18 Feb 2026
- FDA inspects the Melrose Park site.
- 11 Mar 2026
- 483 response filed.
- 16 Jul 2026
- Recalls initiated following FDA discussion on 9 July.
- 22 Sep 2026
- Warning letter. Independent assessments and a three-year retrospective review of critical investigations required.
What the firm said
This is the response most firms write, and it contains real work. The firm acknowledged, in FDA’s words, that “some interventions did not simulate the maximum risk to your aseptic processes.” It committed to review batch intervention data so future simulations reflect worst cases, to engage a third-party consultant on aseptic oversight, and to revise endotoxin sampling when gram-negative organisms appear in bioburden.
“you commit to implement an intervention docking and tracking system and enhanced oversight of related data.” FDA, summarising the firm’s 483 response
Each is a reasonable quality-system improvement. The difficulty is what they have in common: every one changes what the firm observes about the line, and none changes the line.
Why FDA rejected it
A better simulation is still a simulation of the same line
FDA acknowledged the commitment to appropriate media fills, then pointed out that the response did not address sufficient batch sizes and lacked detail on tracking interventions. Then the central point:
“Notably, media fills cannot justify unsuitable manufacturing lines and processing conditions.” FDA, Warning Letter 320-26-129
The design flaws had no dates
FDA found the response “does not provide clear timelines for addressing the fundamental design flaws” of the limited-barrier lines, which it described as “inherently vulnerable to excessive contamination risk.” The letter names layout, ergonomics, excessive manipulations and insufficient barrier separation. A commitment without a date reads as a decision not yet taken.
Containment followed the failure, not the mechanism
After the media fill failures, the firm held batches on the affected lines. It did not hold liquid products on limited-barrier lines that, per FDA, “shared operators from the processing lines associated with deficient aseptic techniques.” If the mechanism is operators on a line that demands excessive interventions, the hold belongs wherever that mechanism operates.
A passing release test answered a bioburden signal
At disposition the firm “did not consider the excessive bioburden to constitute an elevated endotoxin risk.” FDA’s position: “Any batch testing that reveals atypical bioburden levels is a signal that necessitates prompt attention.” Passing finished-product tests were the evidence offered.
The pattern worth carrying forward
Media fills, release tests and observer programmes share a property: they sample the output of a process. They can show that a line failed, but they cannot make a poorly barriered line well barriered. When a response leads with improved sampling, it quietly assumes the process underneath is sound enough to be sampled into compliance.
A tracking system tells you how many interventions the line needs, and a high number is the argument for changing the line.
- Does the action change the line, or only how you observe it? Tracking, trending, enhanced sampling and improved simulations are measurement. If the response contains no change to layout, barrier, flow or interventions, FDA will read it as a decision to keep the design.
- Is there a date on every physical or design change? If a redesign needs a capital project, say so with milestones and the interim risk controls that apply until it lands. An undated commitment is graded as absent.
- Does your containment follow the mechanism or the batch? If the cause is shared across operators, rooms or lines, the hold, the review and the risk assessment must reach everywhere it is shared, not only the line where it was caught.
A defensible response would have kept the measurement commitments and put a second column beside them: a dated plan for the limited-barrier lines, interim controls for the period before it completes, the products and lines sharing operators held or assessed, and a documented view of why gram-negative bioburden at action level is an endotoxin question regardless of the release result.
A passing test shows how the line performed on the day. It does not show that the line is suitable.
Veritas Quality Consultants works with pharmaceutical, medical device, and biologics manufacturers on Form 483 responses, root cause analysis, and CAPA development — built to FDA expectations inside the 15-day window.
Request a consultation