From Form 483 to Import Alert in Six Months: The Pharmathen Case

Plenty of companies pick up a 483 and never hear about it again. The GMP inspector writes down what they saw before leaving the site, the firm sends a credible response, fixes the underlying problems, and the file "quietly" closes. Pharmathen International went the other way.

In a little over six months, that November 2025 inspection became a US FDA Warning Letter, an Import Alert that blocks everything the site makes from entering the United States, two (voluntary) recalls, and an open-ended shutdown of all US production.

The interesting part is how fast it moved, so it is worth exploring what happened and why the US FDA was so strict, and we wholly agree with their verdict.

The site and the inspection

Pharmathen is an established Greek developer and contract manufacturer, best known for complex generics and long-acting injectables. The site fills sterile products for the US market. Two US FDA investigators at Pharmethen from 10 to 21 November 2025, and left nine observations.

Two weeks is a long time for an inspection, and that on its own tells you something.

The FDA’s Investigations Operations Manual lets inspectors stay longer whenever what they find justifies it. A routine drug GMP inspection might be scheduled for three to five days, but sterile operations, injectable products and a site with prior history all earn more attention.

Foreign inspections tend to run longer still, partly because the FDA sends its investigators on multi-week trips rather than the short domestic visits they can do closer to home. So a ten day inspection at a Greek sterile site with problems piling up is no surprise. By the time they had stayed that long, they had clearly found plenty of reasons to.

The observations covered the core of sterility assurance: aseptic technique, facility design, airflow, environmental monitoring, how the firm investigated its own failures, and whether its records could be trusted.

The Warning Letter cites four CGMP regulations (21 CFR 211.113(b), 211.192, 211.42(c)(10) and 211.160(a)) and concludes that the firm’s drug products are adulterated under the FD&C Act.

What the investigators found

Taken together, the observations describe a sterile operation that had lost control of the environment.

Aseptic techniques

Operators were seen reaching over open vials and sterile surfaces, heads inside the filling barrier, hands passing over product-contact surfaces. Forceps were carried from Grade A into Grade B and back without being disinfected. One operator wore goggles with a hole in the top of the frame. Covers taken off sterilised equipment were thrown on the floor, picked up, stored in a cabinet and reused, with no procedure saying whether that was allowed or how often. 

Facility design and upkeep

To start, the pressure regime was backwards. 

The aseptic filling rooms sat at lower pressure than the Grade B corridors next door, which is the opposite of what you want when the whole point is to push contamination away from the product.

The firm’s own mould investigation had already traced the problem to air being drawn in from the technical space because of that. The instruments for differential pressure, temperature and humidity showed live readings only, with no history and no alarm record, so any excursion that happened while nobody was watching went undocumented. This breaches some data integrity principles too.

A HEPA membrane distributor above the filling and stoppering area was simply missing, and nobody had assessed what that did to the airflow. And in the corridor feeding the sterile area's: leaking, discoloured ceiling tiles, a blackish mould-like substance on the walls, ceilings and cold-room fans that came back positive for microbial growth, and insects like flies and wasps.

Smoke studies and media fills (APS)

The smoke studies did not fit their intended purpose. Smoke drifted outwards near the top of the filling machine, the static video showed turbulent air above it, and the camera angles and smoke-generator placement made it impossible to judge airflow direction during real operations.

The Warning Letter describes air leaving the line, bouncing off the operator’s chest, and coming back into the filling area. The media fills/aseptic process simulations had their own problems: intervention types and frequencies did not match real production, not every intact vial was incubated, and some vials were rejected from the test for cosmetic reasons.

In other words, the APS was easier than the process it was supposed to reproduce.

Environmental monitoring and microbial identification

Settle plates were routinely put away from the highest-risk spots rather than at them.

The product-contact surfaces that matter most, the stopper bowls, tracks and forceps, were not sampled at all. Plates used in areas sprayed with disinfectant contained no neutraliser, which quietly nudges results toward “clean”.

And organisms found in pre-sterilisation bioburden and around terminally-sterilised product were never identified, so the firm had no way of knowing whether any of them could survive its sterilisation step or throw off endotoxin.

Visual inspection and a 2022 repeat

Someone changed the visual-inspection process without change control, without updating the SOP and without retraining anyone; the new instructions were passed on by word of mouth.

When the tighter checks then turned up foreign particles above the rejection limit, nobody went back to look at in-date batches made under the old method.

The most common particles were identified as API, with no root cause beyond a line about “residues from a complex process”, and blue cellulose and acrylic fibres turned up with no source ever pinned down. The one that really hurt was the visual-inspection programme still having no alternative destructive testing, the same point FDA had raised back in 2022. That kind of repeat tends to harden FDA’s view of everything else in the report.

Data integrity and investigations

Microbiology was running sterility and environmental tests with no documentation written at the time; analysts filled in sampling records during plate reading rather than when they took the sample. A binder of particle-monitoring records held incomplete, unsigned, unreviewed sheets, with sticky notes flagging who still needed to sign.

The IQVIA software used to print GMP documents could produce multiple copies under the same serial number, with no audit trail and no reconciliation between what was printed and what came back to QA. And the investigations themselves never reached root cause: five years of recurring sterility and media-fill failures, a long list of human-source and gram-negative organisms, and corrective actions that never stuck. One out-of-specification assay result was invalidated after a retest, with no real scientific rationale and no look at the manufacturing side.

The response that didn’t land

Pharmathen replied to the 483 on 15 December 2025.

The Warning Letter works through that response observation by observation and keeps reaching the same conclusion: not good enough.

The firm accepted most of the observations but the fixes were thin on root cause, thin on scope, and short of any evidence they would actually work.

In March 2026 the FDA had published its first draft guidance specifically on how to respond to Form 483 observations, and it asks for exactly what was missing from the response: depth, real root-cause analysis, evidence, and a reply inside the 15-working-day window. 

The process from inspection to suspension

Laid out by date.

FDA’s Office of Manufacturing Quality, whose director Francis Godwin signed the letter, also told the firm to bring in a qualified CGMP consultant and to call CDER’s Drug Shortages Staff before doing anything that might interrupt supply. Trade press has since reported that Cipla, which had partnered with Pharmathen on a long-acting lanreotide product, is already looking for other sites to transfer the work to. 

What happens to the MIA of Pharmathen?

The FDA warning letter and import alert bar only the US market. They have no automatic legal effect on the site's EU GMP certificate. As of now there is no published EU Statement of GMP Non-Compliance for the site in the Eudra