Media Fill Deficiencies in FDA 483s and Warning Letters: Common Citations and Expected Corrective Actions

Media fill (aseptic process simulation) deficiencies in FDA 483s and warning letters: what FDA cites and what it expects

Aseptic process simulation (APS), or media fill, is the main evidence that an aseptic line can produce sterile product. FDA Form 483 observations and warning letters issued to sterile drug, biologic, and outsourcing (503B) and compounding facilities return to the same set of failures. Recurring themes include:

1. Simulations that do not represent worst-case commercial production.
2. Poor intervention records, so the firm has no reliable data on which to base the APS design.
3. Integral units excluded or discarded before incubation, and incomplete unit reconciliation.
4. Weak incubation and post-incubation inspection practices.
5. Failed media fills that are not fully investigated, or that are invalidated without justification.
6. Personnel "qualified" without performing their actual aseptic duties in a media fill.
7. Media fills performed too rarely, or lines not requalified after failures or shutdowns.
8. Growth medium not shown to support growth.

When FDA asks for corrections, the requests are fairly consistent. Firms are asked for an independent review of the whole media fill program, a look back at historical media fills and at batches already distributed, rewritten investigation and reconciliation procedures, requalification after a failure (three consecutive successful runs where no clear cause is found), and stronger management oversight.

The baseline FDA measures firms against

FDA's guidance Sterile Drug Products Produced by Aseptic Processing: Current Good Manufacturing Practice is the benchmark that most citations implicitly apply. It recommends:

• Initial and periodic qualification. At least three consecutive, separate successful media fills for initial line qualification, then semiannual requalification of each processing line. Additional media fills should be considered after changes or events that could affect contamination control ³¹².
• Shifts and interventions. Semiannual runs should represent each shift's activities and interventions, including shift changeover. In warning letters, FDA has also objected when a media fill intervention removed more units than the corresponding production intervention ³¹⁴³¹⁵.
• Personnel. Everyone authorized to enter the aseptic processing room during manufacturing, including technicians and maintenance staff, should participate in a media fill at least annually, in a manner consistent with their routine duties ³¹⁴.
• Incubation. Incubate at a temperature suitable for recovering bioburden and environmental isolates, within 20 to 35°C and within ±2.5°C of target. Incubate for at least 14 days; if two temperatures are used, run at least 7 days at each, starting with the lower one ³¹⁶.
• Inspection. Appropriately trained personnel should examine every media-filled unit, with QC oversight, and bring suspect units immediately to a QC microbiologist ³¹⁷.
• Failures. Investigate to determine the origin and scope of contamination, then confirm restored control with further simulations. If no well-supported cause is found, three consecutive successful runs with increased production scrutiny may be warranted ³¹⁹. Recurring contaminated units signal an adverse trend regardless of the numerical acceptance criteria ³¹⁸.

1. Simulations that do not represent worst-case production

This deficiency is cited repeatedly in both warning letters and 483s.

Warning letter examples

• Sun Pharmaceutical Industries: FDA said the firm had not defined or justified the most challenging conditions for each fill, including container size, maximum hold times, personnel present, and routine and non-routine interventions ¹²².
• Firson Co., Ltd.: Commercial fills were longer and more manually intensive than the simulation, which filled fewer units over a shorter period. The APS also did not fully simulate planned and unplanned interventions or the maximum number of people allowed in the room ¹²⁷.
• Preservation Solutions: The protocol did not define routine and non-routine interventions, the longest permitted line run, normal processing duration, staffing and shift changes, or normal production speed ¹²⁵¹²⁶.
• Dr. Reddy's Laboratories: Media fills left out critical manual interventions. FDA called the firm's conclusion that those interventions posed only a remote sterility hazard unjustified ¹²⁸²⁵⁰.
• Lonza Walkersville: The APS procedure did not define non-routine interventions. Some production lots exceeded the simulated intervention and fill times, and the APS did not set how long personnel could remain in the filling area ¹²⁹.
• Celltrion: During a simulated power failure, the firm excluded integral vials. FDA said removing more units than would be cleared in production undermined the worst-case simulation ¹³¹.
• Kilitch Healthcare India: FDA stated that simulated quantity and duration should closely resemble manufacturing ¹³⁰.

Form 483 examples

• Laboratorios Farmaceuticos Rovi: Media fills did not represent routine intervention counts, production duration, holding time, or the maximum number of people present ¹⁸⁹.
• Ameridose: Media fills did not challenge repeated filling from stored stock solution, representative container-closure sizes, or maximum personnel ¹⁹⁰.
• Dr. Reddy's: An intervention was simulated during line setup instead of during aseptic filling ¹⁸¹.
• Shilpa Medicare: The firm could not show that a simulated engineering repair reflected routine conditions or justify its frequency ¹⁸⁸.
• Outsourcing and compounding facilities: Mismatches here are often stark. One firm's operator simulations filled 10 syringes against commercial batches of 500 to 600 ¹⁷⁴. Others did not challenge production quantities, operator fatigue, or prolonged aseptic work ¹⁸³¹⁸⁴.

What FDA expected. A media fill program summary showing how simulations cover worst-case commercial conditions ⁸⁷. An independent review so that media fills accurately simulate commercial operations, including worst-case conditions ³²⁸. CAPA that results in accurate simulation of worst-case conditions ²⁶⁹.

2. Intervention records too poor to design the APS

The APS design is only as good as the production data behind it, and FDA increasingly cites the upstream records.

• Brassica Pharma: Production records routinely left out interventions, line speed, and stoppages. Interventions seen in production were either missing from media fills or simulated at too low a quantity or duration ¹³⁷.
• Sterling Pharmaceutical Services: Non-routine interventions were not recorded, which weakened the firm's ability to simulate their number and type ⁹².
• Altaire Pharmaceuticals: Records lacked a representative number, type, and complexity of interventions, and left out maintenance and equipment adjustments ¹³⁴.
• Nephron SC (483): Intervention records did not consistently supply the data needed to design fills at representative numbers, types, and frequencies ¹⁹¹.
• JHP Pharmaceuticals and Jiangsu Hengrui (483s): Master plans did not specify intervention frequency, or did not capture the start, stop, and total times of interventions ³⁰¹³⁰².

3. Unit reconciliation and exclusion of integral units

Removing filled, integral units before incubation is a classic way to bias an APS result, and FDA treats it as a serious finding.

Units rejected or removed without a documented cause

• Sun Pharma discarded rejected vials without recorded assignable causes ¹⁰¹.
• Hospira Healthcare India rejected 272 vials in one media fill without documented reasons ¹⁰⁶.
• Ranbaxy (Dewas) removed and destroyed filled vials. FDA warned that this could bias results ¹⁰⁴.
• Teva Pharmaceutical Works rejected integral units without adequate justification, including 4,222 rejected vials in one batch ⁶⁹.
• After a conveyor failure, Teva did not incubate 3,696 integral vials from an aborted run ³⁶⁴.

Units removed for other purposes

• Takeda removed integral vials for unrelated analyses ⁶⁸. FDA said this weakened the simulation's ability to detect contamination ²⁶⁰.

Incubation counts that did not reconcile

• At Novartis and Sandoz, fewer vials were evaluated than were received for incubation, and the discrepancies recurred after corrective actions ¹⁰²¹⁰³.
• Ben Venue, Genzyme, Eli Lilly, and Jubilant HollisterStier received 483 observations for unaccounted units or for rejects discarded without an assignable cause ^18262728.

What FDA expected.

• Documentation accounting for every unit through filling, rejection, incubation, positive controls, and final inspection ¹⁰².
• Updated procedures with justification for any unit not incubated ¹⁰⁵.
• At Celltrion, a retrospective assessment of every media fill since January 2014, covering unit and rejection counts, positives, and explanations for removed vials, plus a CAPA governing rejection of media fill units ⁸³⁸⁴.

4. Incubation conditions and inspection of incubated units

• Brassica Pharma: Inspection relied only on color change. FDA said units must also be examined for pellicles, particles, and turbidity ¹⁰⁷.
• Hospira, Inc. (483): The firm did not specify training for personnel inspecting incubated vials for turbidity and particulates ²¹²².
• Panacea Biotec Pharma (483): The process for qualifying microbiologists to inspect media fill containers was deficient ²⁹⁶.
• ProRx (483): Only selected vials were examined for turbidity, with no scientific justification ³⁵.
• Eli Lilly (483): Units were incubated at 20 to 25°C without the justification the firm's own procedure required ²⁸²⁹³⁰.
• Hospira, Inc. (483): The firm did not run anaerobic media fills under anaerobic conditions for products with the described overlay ²¹⁰.

5. Failed media fills: weak investigations and unjustified invalidations

FDA consistently treats a positive unit as a signal of loss of process control, not as an artifact to be explained away.

Weak investigations

• Cipla: The firm counted a contaminated qualification fill as successful by attributing the positive unit to a handling puncture without adequate evidence. FDA also found a pattern of aborted and contaminated fills that had never been evaluated together, alongside recurring gram-negative findings ^249359360.
• Promed Exports: The firm assigned probable causes to two failed simulations without studies to confirm them ²⁴⁶.
• K.C. Pharmaceuticals: Two failure investigations were left incomplete, and a later failure was invalidated without scientific justification ²⁴⁷.
• Abraxis Bioscience: After multiple failures, the firm released potentially affected batches following one successful media fill. Its investigations did not establish root cause or scope of impact ²⁴⁵.

Form 483 examples

• Fusion IV Pharmaceuticals listed failed runs as passing and replaced failed or cancelled runs, as a repeat observation ⁵⁵⁵⁶.
• Intact Pharmaceuticals opened no root cause investigations for multiple failed lots ⁶⁰.
• AnazaoHealth classified positive leaking vials as "No Test" ⁶².
• Sato Pharmaceutical had continuing failures after a Bacillus cereus positive ⁶⁴.
• Catalent Indiana opened no deviation for terminated media fill lots ⁶⁶.

What FDA expected.

• Revised investigation procedures that require identification of isolates and confirmation of root cause ²⁶¹³⁶⁵.
• An independent third-party review of the media fill program, an independent investigation of recurring organisms, and a retrospective evaluation of four years of media fill contamination events, invalidations, and sterility positives ⁸⁸²⁶⁷.
• Comprehensive CAPA followed by three consecutive successful runs before requalification. FDA rejected Nephron's use of a single repeat fill after two Pseudomonas aeruginosa failures ²⁴⁸³²⁴.

6. Personnel qualification that does not reflect actual duties

Warning letters

• Celltrion's procedure did not require everyone authorized to enter aseptic rooms to participate in a media fill at least annually ²³⁵.
• Global Pharma Healthcare could not show that all cleanroom operators had been qualified through media fill participation ²³⁴.
• Merck KGaA's program did not include all aseptic personnel, and records did not show that personnel performed representative interventions ¹⁴⁸.
• At Takeda, not all personnel responsible for a manually intensive step were required to simulate it ¹²³.

Form 483 observations

• Lupin, Sun Pharma, Sentiss, Akorn, and Jiangsu Hengrui treated operators as qualified even though their qualification process did not specify which activities each operator had to perform, or the operators performed no interventions at all. In some cases, operators performed aseptic connections in production that they had never performed in a media fill ^{182285286287288}.
• Genzyme used an alternative to initial media fill participation without documenting why it was comparable ²⁹⁰.

7. Frequency, requalification, and post-event media fills

• Firson and Samson Pharmaceuticals: FDA said media fills should be conducted at least semiannually for each shift and processing line ^323262263.
• Takeda: The firm resumed aseptic filling after a shutdown that compromised cleanroom control without the post-shutdown media fill its own procedure required, then shipped batches. FDA said environmental monitoring and utility data alone could not show that control had been restored ³²⁵³²⁶.
• Eugia Pharma: A requalification left out a critical aseptic connection intervention ³²⁷. A 483 also noted that no media fill had been run since July 2022 for a process that continued to ship to the U.S. ³⁴⁹.
• Compounders (483s): Several were cited for media fills not performed semiannually ³⁴⁷³⁴⁸.

8. Growth promotion and media suitability

Growth promotion findings in this review appear largely in 483s issued to outsourcing facilities and compounders. Typical observations include:

• No growth promotion test on purchased or in-house media ²⁴²¹⁷.
• No positive and negative controls ²¹²²²⁰.
• Media lots used before suitability was shown ²²²²²³.

At Smart Surgical (dba Burst Biologics), FDA found no evidence that the media's growth-promoting ability had been demonstrated. It also found that units were not incubated under conditions sufficient to detect difficult-to-culture organisms ¹⁵⁵.

Biologics-specific observations

Media fill findings at biologic and HCT/P-adjacent manufacturers follow the same pattern, sometimes in more basic form:

• INCELL Corporation: No APS had been performed since manufacturing began in February 2020. FDA rejected sterility test history and complaint monitoring as substitutes for validation ¹⁵²¹⁵³.
• Smart Surgical (Burst Biologics): No media fills had been run before October 31, 2019 ¹⁵⁴.
• Signature Biologics: Media fills did not represent the maximum commercial batch size ¹⁵⁰.
• Lonza Walkersville: FDA found the firm's APS response only partly adequate. It added that citing a successful APS to justify continued filling requires adequate operational, point-of-fill, and per-run personnel monitoring ²⁶⁴²⁶⁵.
• Samsung Biologics (483): The firm did not keep media fill video recordings that its procedure required for management review of aseptic technique ²⁹⁹.

Corrective actions FDA expected

Practical implications for sterile manufacturers

Read together, these citations show that FDA evaluates the APS as a system, not as a single pass/fail run. Inspectors trace a line from production intervention records, to APS protocol design, to unit accountability, to incubation and inspection, and finally to how failures and invalidations are handled. A weakness at any point can undermine the conclusion that the line is in control.

In responses, FDA has repeatedly found a narrow, event-specific fix inadequate. It has expected firms to:

• assess impact on product already in distribution ²⁵⁹²⁶⁸;
• evaluate failures and invalidations as a trend rather than one at a time ²⁶⁷; and
• bring in independent expertise to review the program ⁸⁵⁸⁸.

Firms preparing for inspection should be able to show three things:

• Every intervention seen in production has a simulated counterpart, at representative frequency and duration.
• Every filled integral unit has been incubated, or its exclusion is documented with a specific cause.
• Every positive or aborted run has a scientifically supported investigation.