Merck slammed with 26-item 483 after inspection of vaccine facility.Merck & Co., West Point, PA
An inspection by FDA investigators Rose Ashley and Omotunde Osunsanmi, with assistance from product specialists Willie Vann, Scott Norris, Dino Feigelstock and Clement Meseda, of Merck's vaccine manufacturing facility in West Point, PA, Jan. 27-Feb. 12, 2009, resulted in a 483 detailing 26 separate observations of nonconformities with GMPs.
The FDA team found that the syringe filling line setpoint was changed on Oct. 26, 2007, to a setpoint that was not validated. The root cause of this violation, they noted, was that "a supervisor made a setpoint change outside of Automation Change Control procedures." As a result, several lots of Gardasil and Recombivax were affected by the deviation.
Unauthorized program parameter setpoints on inspection machines were made at an undetermined time, the FDAers stated. Discrepancies were identified for major defects, minor defects and critical defects. "The root cause was inadequate oversight of program parameters and system security limitations," the inspectors wrote. "The firm lacks traceability for who made the changes. Product lots impacted by this deviation were not identified."
Merck also lacked traceability for determining who disabled steam pressure and temperature alarms on a vial filling line, the 483 noted. Nor could the company identify product lots affected by the deviation, which occurred sometime after Jan. 4, 2007.
"The firm observed recurring events of elevated particulate reject rates during automated inspection of lyophilized products," the investigators reported. Deviation investigations had been initiated since Oct. 5, 2006, for stainless steel particles affecting a number of product lots.
"The firm determined the root cause of the stainless steel particles is metal-to-metal contact of the stoppering systems on the filling lines," the 483 stated. "The firm rejects vials with particulates that are rejected by the automated inspection system and which are found to have particulates during manual inspection."
The FDAers stated: "There is no assurance that stainless steel particles are not in released lyophilized product." The automated filling machines on certain filling lines "are not validated to detect particulates inside or on the side of the lyophilized cakes," but only for particles on top of the cakes, the team explained. Therefore, there is no assurance that particles inside the cakes "will be detected and rejected during the inspection of the product."
Further, there can be no assurance "that stainless steel particles in the lyophilized product do not affect the stability of the product. The firm has not evaluated the impact of stainless steel particles on stability," the report noted.
The inspection found that a study relating to the appropriate shipment of Gardasil, Recombivax and Vaqta vials and syringes did not reflect Merck's actual practice in how these products are packed.
"The firm's investigations of complaints in which returned samples had cracked vials were insufficient to determine if the cracked vials were associated with vial defects or vaccine filling, packaging or shipping operations," the investigators reported.
Merck saw an increase in complaints regarding cracked or broken vials of vaccine products manufactured since October 2006. "No root cause was determined for the increased complaint rate," the 483 stated, which affected lots of Gardasil and Recombivax.
Merck concluded, in the case of a complaint regarding one lot of Gardasil, "that these breaks represent the normal random variation of the whole process of providing this product to their customers." However, the team added, Merck did not analyze other vials of Gardasil and Recombivax that were returned due to under-filled vials to determine what type of fracture or failure mode resulted in the cracked vials.
"The Cracked/Broken Vials Complaint Reduction Investigation determined that the associated defects were heel and neck fractures, and the majority of the product cartons were in pristine condition," the FDA team reported. "The root cause of the cracked and broken vials was not determined to be related to the vial manufacturing process and has not been established to date."
Individual investigations were conducted, but Merck did not initiate a "formal comprehensive investigation into complaints received from different health care provider in different geographical areas from January 2008 to January 2009 regarding bubble/foaming of Zostavax vaccine vials upon reconstitution," the inspection report related, nor for the increase in such complaints from 2007 to 2008.
"Justification provided for the lack of comprehensive investigation was that the complaints were not considered critical and was not one of the top five categories for vials and syringes frequency of complaints received which require that assessments be performed."
FDA found the investigations into this problem with Zostavax inadequate and incomplete. Although Merck said that its review of manufacturing documentation revealed no significant observations, the FDA team found "that a Notification of Cease Release Process dated Sept. 20, 2008, was issued for multiple products" that included manufactured bulk and vial lots of Zostavax that had generated complaints.
Merck's Biological Product Deviation Reporting (BPDR) system procedure defined "significant labeling or package insert error (i.e., information found to be incorrect or missing) including product name/type, lot number, storage temperature, administration route, concentration or volume, or expiration date, as a Biological Deviation."
The procedure did not require a BPDR for single reports of missing labels "but requires a BPDR for reports that exceed the labeling/packaging AQL for the complaint product lot." As a result, Merck did not submit BPDRs for four complaints that involved one or two vials in a larger order that were missing labels or whose labels did not contain the required information.
After an inspection in January 2008, Merck corrected an observed deficiency concerning leaks in Gardasil syringes by implementing an algorithm for syringe leaking complaints prior to use. The FDA team found this algorithm deficient because it and the directions for its use were not formalized in an approved procedure; the algorithm requires a set number of leak occurrences associated with con firmed defective syringes before a BPDR is considered, if the initial review of the complaint history for that lot does not identify that the defect occurred during processing; and the firm did not implement an algorithm or any other guidance for evaluating leak vials for BPDR events.
Merck detected contamination in bulk lots of Vaqta during 2007. "The firm estimated (based on production process and testing data) that the contamination will not affect the product and therefore no BPDR was filed." The one bulk placed on stability did not represent the "worst-case scenario" for the contaminated bulks, the FDA team found.
"Evaluation of the impact of changes and deviations on PedvaxHIB product does not always take into full consideration the sensitivity of the manufacturing process to changes," the FDA investigators reported.
The type of seal used on the agitator of a reactor vessel was replaced with another type that improves wear and the ability to maintain a vacuum, the team stated. "This change was not deemed necessary to report to the regulatory agency because the change was thought not to have product impact. This conclusion does not take into consideration that the quality of the vacuum in the distillation step is a critical part of the derivatization process."
An incorrect protein value, the result of an analyst error, was used in calculations for the conjugation through capping steps, the team noted. Although Merck concluded that this error would have no product impact, the FDAers found that "the review and basis for the conclusion did not include the results of release test performed" on a lot of the vaccine.
The inspection found that equipment was not always adequately tested or qualified before being returned to service. For example, a new weight controller used during disfiltration of solutions was installed and tested for operation of the low weight alarm, but not tested for proper control of the weight of the container, the 483 reported. As a result, during a manufacturing process the new controller did not automatically control the weight, causing operators to have to perform manual control of container weight.
"The firm does not trend or otherwise evaluate all vial and syringe glass breaking events during filling operations," the investigators observed. "Glass breakage events are documented in the batch record or a Glass Breakage Monitoring Logbook form which is placed in each individual batch record."
However, deviation alerts or investigations are required only for "unique glass breakage events," which do not occur at known pinch points; improper line clearances in response to glass breakage; breakages that require grouping product; and fragments in final filled containers. "There is no limit established for the number of glass breakage events that can occur during filling and there is no trending of these events to determine when an investigation should be initiated."
Merck's investigations into defective vial components were deficient, the inspection found. The company found a defect that "can critically weaken the vial" in three instances but did not complain to the vendor or initiate an investigation, according to the 483. A Merck operator also identified vials with defects that were caused by "improper adjustment of the punch out burner during vial manufacturing by the vendor, and the inability of the vendor's automated visual inspection system to detect these defects."
Although the vendor implemented corrective actions that were effective for vial lots manufactured after March 31, 2008, Merck "did not implement corrective actions or controls to address vials in inventory that were manufactured prior to implementation of the vendor's corrective actions."
At the time of the inspection, the company still maintained glass vial lots manufactured prior to the vendor's corrective action, the team reported.
The FDA investigators found fault with the timeliness of Merck's corrective actions. "The firm has recurring glass breakage events at the exit star wheel on the vial filling line" in one department "secondary to over-torque of the wheel. A procedure for setting appropriate star wheel torque has not been established to prevent further breaks."
The operation of the fermentation, purification and recovery system and the training level of its operators prior to initiating a production run was found deficient. The team found that leaks during production runs created ruptured gaskets and pump seals.
Visual inspections of incoming vial components were conducted under normal light in the component inspection area, the FDAers observed. "There are no requirements for light intensity for inspection areas and no requirement for periodic evaluation of the light intensity," they wrote. Further, the investigators found no detailed procedures defining the vial inspection process and no "standardized kit of defects" for training employees who visually inspect vial components.
The inspectors noted that a single Merck vial lot contains components "that typically come from more than one vendor production lot" and comprises material "made across multiple vendor manufacturing lines. The firm did not determine if other vial lots were impacted by vendor vial lots with known defects."
Merck had no data to support the expiration date under its use and storage conditions for certain raw materials used in manufacturing Gardasil, the 483 reported. "The firm stores the raw material in the original container at ambient temperature, and the containers are opened and closed multiple times" and exposed to air. "The firm has not conducted quality testing of the raw material under their use and storage conditions."
There was no designated area for storing incoming samples for analysis, the inspection found. Merck stored such samples in a freezer co-mingled with boxes containing pre-analyzed and research samples. There was no designed place to isolate incoming Lot release samples in the freezer.
Controls of environmental manufacturing microbial bioburden levels were deficient, the FDA team stated. "Specifically, several bioburden excursions including TNTC [too numerous too count] microbial excursion levels were noted" and other microbial organisms in the vaccine manufacturing buildings and areas.
The inspection found "microbial action level excursions, including results of TNTC, were noted in the thaw baths used in the thawing of bulk vaccine products," despite the fact that the baths were cleaned after each use. The investigators found "no documentation of cleaning qualification/validations for any of the thaw baths" and also noted that "there is no documentation that formal investigations have been opened to address the high microbial levels" found in these baths.
Documentation was also lacking of evaluations of the effectiveness of disinfectants "used in the sanitization of the manufacturing areas." Merck conducted investigations into the sterility failures of PedvaxHIB and Comvax bulk lots that led to recalls of some marketed vials of these products, but the inspectors found no evidence that the disinfectants were evaluated in the course of these investigations.
A disinfectant effectiveness study conducted in 2002 was conducted only on stainless steel. "The firm currently has a draft and unsigned disinfectant effectiveness study protocol with no documentation of implementation start date."
Particles and impurities were noted in some lots of reagents that Merck received in 2007 and 2008, the 483 noted. "However, the firm continued to manufacture vaccine products with these raw materials with justification that the use of the raw materials has no impact on product quality in the event that similar particles were present in the materials. No released vaccine lots were placed on stability in regard to these raw materials/impurities."
The investigators observed that impurities were found in a sodium chloride reagent lot "on three separate occasions," yet Merck continued to use the material in manufacturing vaccine. "Trend analysis from Feb. 26, 2007, to Feb. 26, 2008, indicated that this was the third extraneous matter atypical event for the lot" in question.
"Although the firm continues to document extraneous particles/impurities in the above received manufacturing components, inadequate corrective and preventive actions were instituted," the inspection team reported. The vendor responded to Merck's concerns "that it has confirmed through prior investigations that the particles found in the components are inherent to the manufacturing and process train for these materials. As such, no additional CAPAs were instituted at the vendor or at the firm."
The FDAers added that Merck had conducted no "for-cause audits" of its suppliers, and there was "no documentation that attempts were made to re-qualify new suppliers that could supply products with no particles/impurities."
There were no validation studies to support Merck's claims that products manufactured using these raw materials--including Pneumovax, Pedvax, Varicella, Zostavax, Mumpsvax and MMR II--were not affected adversely by these impurities.
Further, Merck "indicated that the continued use of these lots was based on lots of sodium chloride reagents and magnesium chloride hexahydrate that passed USP specifications. However, there are no assurances that the impurities in the lots are homogenously distributed and that the tested portions of the lots by QC contained the particles/impurities that were later noted during dispensing of the lots."
The investigators found that not all sampling of compressed gases sterilizing filters used in manufacturing vaccines were representative of actual use conditions. These filters "are not monitored or sampled at the point of use during manufacturing," they wrote.
"In addition, there is no documentation in any of the SOPs ... that specifically states the location for the testing of compressed gases, i.e., before or after the inline filters, to assure that the gases are tested at the same use points during manufacturing."
Further, not all point-of-use sterilizing/tank vent filters for compressed gases had data to justify the number of uses and the inspectors found no documentation of CAPAs for compressed gas excursions.
The FDA team found incomplete a CAPA implemented for the deletion of a missed licensed identity test. The 483 stated that an analyst "was able to permanently remove" the test record of certain sampling. The CAPA instituted to prevent the recurrence of this problem failed to include additions to the software applications that would prevent the laboratory analysts from deleting required licensed sample tests, the investigators noted.
A CAPA implemented regarding a cleaning revalidation failure was found to be incomplete, as well. The FDAers reported that "information contained in the BPDR that was sent to CBER did not include" a previous related event concerning an investigation begun after black particles and oily residue were found in drums of material received from a vendor.
FDA added: "There were no assurances that one cleaning verification/revalidation run" as a result of this finding, and three other revalidation runs, were conducted as they should have been. The cleaning verification study conducted after the CAPA related to black particles "failed acceptance criteria for conductivity for the final rinse," the team wrote. This cleaning verification also failed to include bioburden and/or endotoxin rinse water samples.
The root cause of the failure of this verification study "was attributed to residue from the rinse water retained inside a tilted 'T' connecting the recirculation loop to the transfer line-tank." The reexecution of the cleaning validation study, which was said to meet conductivity specifications, "has the notation of 'NA' for specification in the documentation of cleaning validation results for conductivity." In addition, no deviation was documented and no investigation was conducted into liquid found during the cleaning validation and swabbing sampling of the port.
The 483 further reported that the inline conductivity probe had malfunctioned and in at least once instance showed an incorrect reading of zero for conductivity. "No comparability study was conducted for the inline conductivity probe used during normal cleaning and laboratory conductivity testing performed during the tank's cleaning validation," the inspectors wrote.
Sampling of water for injection (WFI) was "not representative of production/manufacturing operations use conditions and the sampling of WFI use points is not consistently applied throughout the production areas," the investigation found. Investigations and controls over pre-treatment water for still water production were inadequate to assure the quality of the still water, the team also reported, noting that various microorganisms were noted in pretreatment water during the period February 2008 February 2009.
In addition, a CAPA instituted as a result of WFI endotoxin levels was inadequate. According to the internal investigation, the root causes for these endotoxin excursions were contamination caused by trapped water in the piping panel and excessive feed water into the still. However, Merck failed to include in the CAPA the examination of all WFI pipes.