Liverpool, birthplace of the Beatles, is now also indelibly fixed in memory as the site of a pharmaceutical manufacturing fiasco, which came to light last October when British regulators halted operations at Chiron Corp.'s Fluvirin flu vaccine manufacturing plant.
The story of what went wrong at the plant is complex, one that involves not just Chiron but multiple firms and regulatory bodies. It's also a story of risk analysis and political pressure, thin profit margins, short production timeframes, misplaced optimism and bad timing.
Pharmaceutical Manufacturing has obtained uncensored copies of a 1999 FDA site inspection report and Warning Letter, as well as redacted copies of subsequent inspection reports, all of which suggest that Chiron may have inherited some serious systemic problems when it bought the Liverpool plant in 2003. Some of these problems may have been impossible for Chiron to fix given the amount of time it had to prepare for the 2004-2005 vaccine season.
In the absence of comment from Chiron or FDA, this article will discuss the lessons that drug manufacturing professionals and regulators can learn from this case, and consider its future implications for vaccine manufacturing in the U.S. Within the past few decades, regulators have continued to raise the bar for quality in a risky, low-margin business, and, since the 1970s, the number of vaccine manufacturers in the U.S. has dropped from 25 to 5. Chiron’s case has already promoted frank discussions. We can only hope that it will also lead to the decisive actions required to rebuild the U.S. vaccine supply base.
The story began last summer, when Chiron Corp. discovered gram-negative Serratia microbes in one lot of Fluvirin vaccine manufactured at its U.K. plant. In September, inspectors from Britain’s FDA equivalent, the Medicines and Healthcare Products Regulatory Agency (MHRA), inspected the plant. They found that levels of microbial contamination there had been increasing since March, and that the levels reached in 2004 were “several orders of magnitude” higher than they had been in 2002 and 2003. Gram-negative bacteria were found in the general plant environment, in a sterilizing filtration room, in bulk and finished product.
Chiron, perhaps confident that the source of contamination lay in its upstream manufacturing operations, informed the FDA and CDC that it would be able to meet its supply schedule, only to find days later that MHRA suspended the plant’s manufacturing license. Weeks later, FDA inspectors agreed that the safety of Fluvirin made at the plant could not be guaranteed. Roughly 48 million doses of flu vaccine, nearly half the U.S. supply, could not be shipped. On December 10, FDA issued Chiron a warning letter.
Chiron has been working around the clock to resolve manufacturing issues at the plant. In December, U.K. authorities extended the license suspension, to allow the company more time to carry out remediation efforts. The suspension can be lifted at any time, once the facility demonstrates compliance. Nevertheless, the damage has been done, to Chiron itself and the security of the public health.
“I can’t say that Chiron didn’t do enough due diligence — due diligence can be extremely hard to do — or that FDA inadequately reviewed the manufacturing process,” says Jim Akers, president of Akers-Kennedy and Associates, Inc. (Kansas City, Mo.) and a specialist in aseptic processing. “Part of the problem, no doubt, reflects the continued reliance on older technologies to manufacture vaccines. Some of these technologies should have been eliminated years ago.”
Whatever their source, problems at Chiron’s Liverpool plant have had severe repercussions for the company. After committing itself to a risky market that most U.S. producers have abandoned, Chiron has had to write off $91 million in losses. Its stock price has plummeted. Chiron has also been sued by stockholders, who allege that the company covered up problems at the plant. Meanwhile, a number of opportunistic consumer class-action lawsuits have been filed across the country.
A cynic might wonder why Chiron, or any company, would choose to enter the U.S. flu vaccine market in the first place. Others might question why a savvy biotech firm would invest hundreds of millions of dollars in an old production facility with a long, documented history of systemic quality and GMP problems. “While reading the 1999 FDA inspection reports and warning letters, Chiron’s management team should have found the hairs on the backs of their necks standing up straight,” says Donald Gerson, president of the cGMP and vaccine manufacturing consulting firm Axenic, Inc. (Montebello, N.Y.).
In May 2003, Chiron first announced its intent to buy PowderJect’s flu vaccine business; the Emeryville, Calif.-based firm was to become one of the two leading U.S. flu vaccine producers in a market that had shrunken to three suppliers in less than two decades.
One unmistakable sign of trouble was the fact that the Liverpool plant had changed hands several times in the three years before Chiron bought it: Until 1999, Medeva Pharma owned the facility. The company then sold it to Cell Tech PLC in January 2000. PowderJect bought the facility just seven months later, spending over £85 million on a new filling plant, training programs and other improvements before June 9, 2003, when Chiron officially bought PowderJect for $878 million.
|A Cursed Facility?
When owned by Medeva, the Liverpool vaccine manufacturing plant was plagued by serious quality problems. Four years ago, oral polio vaccine made at the facility was recalled when it was found that the plant might have used raw materials that could have been derived from animals infected with bovine spongiform encephalopathy (BSE) or “Mad Cow.”
In addition, in the U.K., The Observer reported that manufacturing problems in 2000 led to shortages of yellow fever and TB vaccine product from the facility.
To make matters worse, the facility did not establish nor maintain bioburden limits or determine root causes of product contamination. It had no system for validating filtration and other key procedures, while SOP standards and process monitoring were inadequate.
Medeva responded to these issues, announcing new process and other validation projects. FDA allowed the vaccine, which had passed product-release tests, to be sold.
Four years later, on June 2, 2003, one week before ownership of the facility was to be officially transferred to Chiron, FDA inspectors launched another site inspection that revealed ongoing quality issues. FDA found that root causes for prior cases of contamination had not been investigated; vaccine lots with high levels of contamination were refiltered, yet no procedures were in place to determine the stability of these reprocessed batches.
While expanding production to meet 2003-2004 vaccine manufacturing deadlines, could Chiron have, realistically, fixed these problems, or would it have been better simply not to buy the plant? Buying an older facility was a sensible choice, since a new grass roots facility would have taken years and been expensive to build, says Mike Williams, a flu vaccine expert and former FDA inspector who now works for the Biologics Consulting Group, LLC (Alexandria, Va.).
And the facility’s age may not have been the issue. After all, as part of its acquisition of PowderJect, Chiron had also bought an older flu vaccine plant in Marburg, Germany, which has had no quality problems.
In the long term, though, a new plant might have saved Chiron both money and prestige. Even a pristine new plant can have sterilization problems, but, with an old plant, risks increase. “With an existing plant, you want to keep it producing, it’s harder to fix problems and there are limits to what you can do,” says Jim Agalloco, president of the consulting firm Agalloco & Associates, Inc. (Belle Mead, N.J.) and a specialist in validation and aseptic manufacturing. “Money would have been better spent on building brand new facilities from the ground up,” he says.
“I certainly wouldn’t have bought this facility — either not bought it or expected to start all over again from scratch,” says Axenic’s Gerson. Contamination problems may have been extremely difficult to fix, he says. “My guess is that they had good people and a bad facility coupled with production and economic pressures. No doubt they tried hard.”
No root cause analysis, weak validation
Were the plant’s owners negligent in operating the facility? Evidence strongly suggests that problems originating during Medeva’s ownership were passed on to the facility’s subsequent owners.
The most fundamental problem was a failure to conduct a thorough root-cause analysis for aseptic processing problems, says Agalloco. Although contamination at the plant had been traced to a gram-negative organism associated with water, Medeva hadn’t taken the steps needed to remove it from the plant’s water supply. This problem remained unresolved when the plant was owned by PowderJect.
Validation was also weak to nonexistent for several critical functions. FDA documents from 1999 indicate that whatever validation had been done at the facility was done seven years earlier. The data were old, probably not as rigorous as they should have been, and might not have reflected the way that the facility was later run, Agalloco says.
Five years ago, Liverpool plant management could not ensure that reprocessed batches met standards. This in itself was an admission of failure, Axenic’s Gerson says. “The whole idea behind validation is to prove that you can reproducibly do what you say you’re doing,” he says. “If you reprocess with the same undefined method, have you defined the process any more clearly?”
For example, when it was run by Medeva, the facility had no process for validating the concentration of thimerosol, a mercury-derived preservative, in fills. “Thimerosol was the one chemical safeguard they had against contamination and they didn’t prove it was there in the right concentration,” says Gerson. “If the product depends on the process, the process hasn’t proven to be sterile or clean, and the right concentration of thimerosol was not determined, where was the assurance that they were manufacturing safely?”
FDA’s observations don’t indicate product nonconformance, although they do suggest that Medeva could have been more proactive about quality issues, says Peter Weiland, president of the Georgia-based GMP consulting firm QCT Solutions, LLC. Nevertheless, failure to comply with GMPs technically renders a product “adulterated” regardless of testing results.
“The term 'adulterated’ is not always black or white,” he says. FDA determines public risk based on inspectors’ observations, but weighs in such factors as the firm’s willingness to take immediate corrective action, provide periodic updates and alert the Agency of further non-compliance.
The sterile filtration and blending processing steps of monovalent pool and trivalent bulk should have been requalified annually, Weiland says. In addition, reprocessing should have been limited to one time only, and the frequency for analyzing bioburden data clearly identified.
Pareto charts could have been useful for analyzing deviations, he says, to facilitate corrective and preventive actions (CAPA), while historical data could have been reviewed to justify monitoring clean steam and compressed air, so critical to the process.
FDA: At fault? Or between a rock and a hard place?
Some have questioned why FDA allowed vaccine manufacturing to continue at the facility after the Agency’s 1999 inspection.
Within the broad term “bioburden,” endotoxins, or the remains of dead bacteria, posed the greatest risk to product, and to public safety, says Axenic’s Gerson. Endotoxins pose particular risks because there are thousands of different types, and no way of knowing how each will affect each person’s immune system.
“The U.S. has cherished this idea that vaccines should be cheap. Well, why should I protect you from a life-threatening disease — for tetanus, for a lifetime, for the flu, for one year — for the price of a Big Mac or two?”– Don Gerson, president,
In fact, last June, FDA found that PowderJect had not even studied adverse reports linked to specific product lots or attempted to correlate product problems to manufacturing issues. The Agency’s report noted that two healthcare facilities that bought vaccine made at the plant had reported 41 cases of adverse events in patients, all of which could be traced to one batch of Fluvirin made at the facility between 2002 and 2003.
“In the earlier investigation reports, the facility had provided no justification that the levels of bioburden were acceptable,” says Jerry Martin, senior vice president and global technical director of Pall Life Sciences (East Hills, N.Y.) and coauthor of PDA’s Technical Report 26 on Sterilizing Filtration of Liquids. “It is important to validate cleaning cycles for ultrafiltration equipment and to validate that the sterilizing filter is capable of removing bacteria.”
The cold hard fact is that, five years ago, Medeva had adulterated product on its hands, Gerson says. “FDA is generally tough and correct. I don’t know what happened here,” he says.
Other observers disagree. “FDA made a good faith assumption that Medeva had responded to findings and was fixing the problems, says Alan Waldman, president of Waldman Biomedical Consultancy, Inc. (Oceanside, N.Y.).
What would have happened if the Agency had pulled the plant’s manufacturing license last year and refused to accept half the U.S. flu vaccine product? “It would have been a political disaster,” says Agalloco. “It might have been different with a product that could be substituted for. With flu vaccine, the Agency was between a rock and a hard place.”
Risk assessment and due diligence
The June inspection occurred during transfer of plant ownership, so Chiron was clearly at a disadvantage. One can only assume that PowderJect was open in revealing all the facility’s past problems.
There could have been “people problems” involved, not unusual in a plant that has changed hands so many times. Clearly, says Gerson, whoever was in charge of the plant from the late 1990s on couldn’t get senior management to help them solve the facility’s manufacturing problems.
Nevertheless, Chiron could have done a better job at risk assessment and due diligence, says Waldman. In a caveat emptor world, it’s up to the buyer of any facility to do the due diligence. “Chiron appears to have been a somewhat naïve purchaser,” Waldman says. “The company needed a more complete review of the validation master plan, including a much stronger Risk Assessment.”
Complex, primitive process
No doubt the inherent risk and complexity of vaccine manufacturing also led to some of the issues at the facility. “Vaccines are not like chemical entities that can be assayed,” says Gerson. “Often you don’t know what the active ingredient is, or what precisely drives the process.
“As a result, there’s a major ongoing effort where regulators say 'the product is the process,’ and you’re stuck proving — no easy feat — that the process is the same as when you did your original clinical trial, and that it remains the same, batch to batch,” he says. And, if vaccines represent the worst case of this kind of complexity, the flu is the worst case within vaccines.
“Flu vaccine manufacturing is far more complex than production processes for small molecules,” agrees the Biologics Consulting Group’s Williams. Strains change every year, so the product changes every year. Yields change and manufacturers have a very short time in which to get everything ready.
Regulators with training in modern biotechnology haven’t always seemed to understand the complexities of vaccine manufacturing, says former inspector Williams, particularly as CBER has evolved to operate more like CDER. “You have an old manufacturing technology clashing with new guidance from regulators,” he says.
“Many vaccine manufacturing processes were developed 30 to 40 years ago.” says Hélène Pora, vaccine application development director at Pall Life Sciences (East Hills, N.Y.), “The first flu vaccine process was developed in 1937.”
The situation parallels what happened with plasma-derived therapeutic proteins in the late 1990s, says Pall’s Martin. “Team Biologics issued several 483s essentially saying 'you need to operate under current GMPs,’ and manufacturers had to scramble,” he says.
Before he left FDA, Williams says, he spent a lot of time trying to convince inspectors that a certain amount of contamination is inevitable in egg-based vaccine making — if it happens upstream, it can be fixed, but if it occurs in the filling line, it’s a strict cGMP violation.
While regulators need to be aware of the unique issues involved in vaccine manufacturing, increased scrutiny of biologics is a good thing, says Gerson. “The image of the super-tough FDA has not always been true for vaccines,” he says, recalling his experience 10 years ago, when he left a job as head of manufacturing for a Canadian plant to start a similar position in the U.S. The Toronto plant had been overseen by Canadian health authorities, who were fairly demanding.
Gerson was surprised by one relatively lax inspection by FDA. He is pleased that Team Biologics has raised its standards. “If regulators don’t place the scratch on the wall high enough, economics will take over.”
“I have no sympathies for lax standards,” Gerson continues. “Doing vaccine manufacturing right does not cost a whole lot more than doing it wrong. Companies have to look closely at the cost of multiple product failures and throwing batches out. It’s a false economy to have a low quality target.”
Plainly, flu vaccine manufacturers have to clear a number of significant hurdles that aren’t there for their peers in biopharma and pharma. First, there is egg handling and injection. “Inside the egg, conditions may initially be sterile, and you can sanitize the outside of the egg, but once you remove the allantoic fluid it becomes difficult to keep equipment very clean, and to prevent bacterial growth,” says Pall’s Martin.
“Dealing with biological materials is difficult,” says Akers. “Even if you maintain control over colonies providing the eggs, it’s hard to imagine that raw eggs could be produced in a truly aseptic manner. You need a way to control bioburden and harvesting that demands a very high level of contamination control and attention to detail.” Some companies have automated parts of this process.
Such issues are not a problem with traditional pharmaceuticals, which can use classical terminal sterilization techniques to reduce the likelihood of non-sterile product to one in a million, or even a billion.
The bacterial contamination rate of aseptically processed biologics has been steadily improving. In the best facilities, it might approach that of terminal sterilization, while, at the other end of the spectrum, it may be only slightly better than one in a thousand, says Agalloco. “Media fills don’t truly support the contamination rate of any batch, so we have no accurate means to assess this,” he says. “In addition, sterility testing is imperfect at best—a low percentage of contamination in a lot can easily be missed due to the limited sample size.”
Getting more scientific
Since human handling is the root cause of many contamination problems, the best way to eliminate aseptic breaches in vaccine production is to develop more robust processes that don’t require human intervention, says Akers. “Some people act like you can validate human intervention, when you can’t,” he says. “It’s performance-based and depends on the skills of the individual operator, which are extremely hard to control.”
Single-use filtration systems (Pharmaceutical Manufacturing, April/May 2004, p. 34) offer one option, and are attracting more vaccine manufacturers, Pall’s Pora says. FDA inspection documents at the facility had also pointed to the need for better approaches to aseptic connections, an area where the Agency has promised better guidance. Recently, manufacturers have begun to offer preassembled and presterilized filtration systems with improved aseptic connector configurations to minimize handling.
Akers-Kennedy and Associates
Although FDA’s recent Aseptic Processing guidance may change future perceptions, people running plants now were guided by the Agency’s position of three years ago, and few invested in isolators, Agalloco says.
“Only the September 2004 guidance from FDA addresses the isolator issue, and then only in an extremely neutral way,” he says. “The guidance also raises a number of unique issues, though. If you were deciding what to buy today, you might not want to invest in the technology given the ambiguity.”
Part of the problem, Akers says, is that in the past, isolator vendors had overreached, promising perfection before it could be delivered. “FDA’s study was based on real observations and comments from industry,” he says.
ISO is now developing global standards for isolators and for aseptic manufacturing in general. However, if FDA doesn’t make it clear what it wants and accepts, says Agalloco, industry will be slow to adopt isolator technology. “In the U.S., we’ve already fallen behind in adopting newer aseptic processing technology,” he says.
The promise of cell culture
Cell culture technology, and the ability to harvest viruses in mammalian cells, also offers hope of making flu vaccine manufacturing easier to control. It would also eliminate the need for thimerosol, the use of which in pediatric and other vaccines has become a controversial issue of late. However, cell culture technology also has its limitations. “Right now, we still guesstimate what is needed for flu vaccine,” Agalloco says. “It’s not clear that fermentation processes can work in such a short timeframe, or produce all the different strains of virus required.”
Cell culture technologies have been in development for many years, says Biologics Consulting Group’s Williams, and have suffered from relatively poor yields. However, improved manufacturing technologies have led to better yields, and several companies have obtained cell-based manufacturing licenses in Europe and are pursuing licensure in the U.S. “Although these second-generation vaccines may not solve all the problems facing flu vaccine manufacturers, they will allow producers to get away from the problems associated with egg-based vaccines and will definitely make for a better-defined manufacturing process,” says Williams.
The root of all evil
At the root of many of ongoing flu vaccine quality problems, experts agree, is low profitability. “The level of pricing that the government reimburses for flu vaccine is abysmal,” says Agalloco. “You can’t support this price for product. How can companies invest in technology, given the threat of litigation for any adverse effects?”
“The U.S. has cherished this idea that vaccines should be cheap,” says Gerson. “Well, why should I protect you from a life-threatening disease — for tetanus, for a lifetime, for the flu for one year — for the price of a Big Mac or two?” he asks.
Consider the high cost of incorporating airbags or seat belts into cars, which the public accepted, he says. “Companies would comply if they thought they could make a profit from their product,” he adds. “I’ve watched from near and far as vaccine makers have left the U.S. and it’s a tragedy.”
Worsening the economic impact is the fact that whatever vaccine is not used in any given year must be destroyed, and consumption can vary, based on whether physicians promote flu vaccine to their patients. In 2002, 12 million doses of flu vaccine had to be scrapped, according to Howard Pien, Chiron’s president and CEO.
Recently, Dr. Anthony Fauci, director of the National Institute of Allergy and Infectious Diseases at the National Institutes of Health, scoped out some suggestions. The government could guarantee the purchase of unused vaccines, he argues. He also suggests that companies could be offered regulatory relief or that manufacturer liability could be limited.
Although these points appear to make sense, they would have to be addressed very carefully. Aseptic processing, and vaccine manufacturing, needs to apply the highest standards to quality and not simply accept the status quo, says Akers. “The industry needs to achieve world-class levels of process knowledge.”
At this point, the industry clearly isn’t there yet, and vaccine manufacturing — and particularly flu vaccine making — remains an imperfect world. Both Chiron and FDA did the best they could. It just wasn’t good enough.
|Problems Persist Four Years Later
In 2003, a different team of FDA investigators found the following:
|Timeline for a Troubled Plant
1999 – FDA inspectors find major quality systems problems at Medeva’s Liverpool flu vaccine manufacturing plant, including mixing of contaminated and uncontaminated monovalent lots. Regulators had previously issued a four-point 483 letter to the plant outlining problems with use of connectors and tubing.
January 2000 – Medeva Pharma sells the plant to Celltech.
July 2000 – Celltech sells the facility to PowderJect, which spends £85 million on additional modernization and training programs.
June 2-9, 2003 – FDA inspections reveal ongoing quality problems at the facility, including refiltering of contaminated lots without stability assessments, and failure to adequately determine potential root causes of contamination and to develop corrective actions.
June 10, 2003 – Chiron finalizes its acquisition of PowderJect, including its U.K. facility. Over the next six months, Chiron increases flu vaccine manufacturing capacity at the plant by 50%, to 38 million doses. Chiron also announces plans to build a new filling facility at the site, to expand egg-based vaccine manufacturing, and to deploy cell culture vaccine production after Phase III studies the following year. Plans for a $200 million cell culture plant in Germany are disclosed.
August 2004 – Chiron discovers bacterial contamination in about 4 million doses of Fluvirin, quarantining product pending an investigation.
September 2004 – The U.K. regulatory agency, MHRA, issues warning letter to Chiron, requiring that product be quarantined.
September 28, 2004 – Chiron management assures U.S. regulators that the company will meet its supply quota in October and that it has identified the root cause for contamination and believes it to be limited. Testing on representative lots of finished product show vaccine to be free of contamination; CDC and FDA arrange to buy additional vaccine from other suppliers.
– MHRA inspects the facility.
September 30, 2004 – MHRA releases initial findings of its inspection, which found that, since March 2004, bioburden at the facility had increased by several orders of magnitude compared with 2002 and 2003. Instances of gram-negative bacterial contamination had increased in a sterile filration room, and the organisms were also found in the environment, in monovalent bulk product and finished product vials.
October 4, 2004 – After completing its investigation, MHRA decides to suspend Chiron’s manufacturing license.
October 5 to 10, 2004 – FDA team meets with MHRA, senior leadership of Chiron.
October 10, 2004 – FDA inspects the plant.
October 15, 2004 – FDA completes investigation, finding quality problems and concerns. Despite retesting and negative test results for product contamination, the Agency determines that problems could be related to filling operations, and that it can’t guarantee product safety.
December 10, 2004 – FDA issues warning letter detailing the plant’s non-compliance issues.