Glass Quality Crisis Prompts Multi-Faceted Array of Risk-Based Improvements

Drug makers, glass suppliers and FDA officials who attended a recent Parenteral Drug Association glass quality conference made some progress toward understanding why glass-related recalls have surged and what corrective and preventive measures they should apply.

PDA organized the May 23-24 gathering in the wake of a sharp increase in glass-related drug recalls over the past year that has contributed to critical drug shortages ("Another Busy Year of Drug Recalls Led to Shortages of Injectables" — "The Gold Sheet," May 2011).

Discussions at the filled-to-capacity event in Arlington, Va., were lively and intense, with moderators periodically intervening to keep things from devolving into finger pointing.

One question that kept resurfacing was about the lone glass defect. Would there be trouble with regulators if a company was doing all it could to avoid marketing vials or syringes that were cracked or contained particulates, but nevertheless received a customer complaint about one? Would FDA expect a recall even if there was just one complaint?

There was a suspicion among some attendees that FDA is moving toward a zero defects stance, and that this is behind the recent increase in glass-related recalls and warning letters.

When the issue erupted again in the closing session, Martin VanTrieste, senior VP of quality at Amgen and conference co-chair, acknowledged “this fear that the FDA expectation is zero defects” when it comes to cracks and particles in glass containers. “It might be a great aspiration,” he said. “But the process technology and the process capabilities, we’ve heard it over the last two days, isn’t there to produce zero defects.”

Rather than directly confirm or deny the suspicions, FDA participants in the discussion emphasized the importance of using risk-based approaches to reduce defect rates as much as possible.

David Jaworski, an FDA compliance official, did say it appears to him that industry is noticing glass defects more, and that it may be experiencing more of them.

“You do see that there is more of an awareness for certain types of defects on the part of companies. They are reporting what they are finding more often, and so there is a shift in that perspective about awareness, and being able to detect,” he said.

Jaworski added that “on the manufacturing side, there are also some shifts going on there that we’re seeing from products. Some people are moving them to different plants. Some of the products are going generic. We do see some changes in the manufacturing processes also that are affecting the products. I can’t get into specifics.”

The one answer that knit together the many discrete issues that comprise the glass quality conundrum was the idea of developing a more sophisticated, more comprehensive risk basis for developing and producing drug products and their glass containers – and for dealing with complaints about defects.

One of the outcomes was for PDA to share with attendees the April 2010 product of a joint industry and regulatory work group in France on how to make decisions on recalls when manufacturers receive complaints about glass particulates. The 12-page document includes an English translation.

A conference participant explained that the French drug regulatory agency, the Agence Francaise de Securite Sanitaire des Produits de Sante, or AFSSAPS, had been taking an “extremely conservative reaction to glass particles. It used to be a single identification of a glass particle in a customer complaint meant a full recall of the lot.”

Often, she said, the World Health Organization would expect to see the French recall expanded globally.

“So they developed this risk assessment that’s intended, much as we talked about a lot today, about working all the way through all the controls in your process to say, ‘is this something that’s a significant risk, or is it possibly maybe a one-off kind of situation,’ so part of this was intended to preclude those instant recall reactions that used to occur.”

She suggested that PDA might want to explore working with FDA to produce a similar document. VanTrieste indicated PDA would likely follow up on that idea with the appropriate agency authorities.

Approximate number of sterile drug product units recalled for glass issues.Source: Joyce Bloomfield, Merck, Sharp & Dohme, based on FDA enforcement reports

An alarming trend in recalls

Joyce Bloomfield, executive director at Merck Sharp & Dohme for Global GMP Systems & Compliance, started off the conference by sharing some sobering statistics on glass quality.

Bloomfield tallied recalls and warning letters in five-year increments to underscore the alarming trend that gave rise to the meeting, which she co-chaired with VanTrieste.

“In 2006 to 2011, there were at least 19 recalls dealing with glass issues, and most of those happened in the last 15 months,” she said. “Those are the events that led us here today. There have also been a number of FDA warning letters. … As a result, we have work to do out of this conference.”

Glass-related recalls jumped from three in 1996-2000 and four in 2001-2005 to 17 in 2006-2010, plus two more in reported early 2011.

Similarly, glass-related warning letters increased from four in 1996-2000 and three in 2001-2005 to nine in 2006-2010.

But the most stunning figure was the increase in the number of units of sterile drug product recalled because of glass issues: from some 395,000 in 1996-2000 and 1.6 million in 2001-2005 to more than 100 million in 2006-2010.

That’s 60 times more than five years before and 250 times more than 10 years before.

Bloomfield went on to show that the types of glass problems also have changed. Glass defects and glass flakes generated very few recalls in 1996-2005. The only glass-related recalls in those years were in the other two categories: cracked glass and glass particulates.

Of late, the main types of containers involved in sterile drug recalls were vials rather than syringes or cartridges, she said, noting that 15 of the 19 2005-2011 recalls involved vials. But she added that there have since been several syringe recalls.

The U.S. is not the only country to experience such recalls, Bloomberg noted. She showed data culled from FDA enforcement reports and announcements of recalls in the U.S., the United Kingdom and Canada, which showed increases in glass-related recalls in the U.K. and Canada, as well as in glass-related U.S. recalls that extended to other countries.

“We really wanted to show that there’s an international trend in recalls on sterile glass products,” she said. “If we could get that data from other countries, I’m sure it would be much more than that.”

The nine glass-related warning letters FDA issued over the past five years addressed biological product deviation investigation filing practices, as well as investigations into glass issues.

There were cases where FDA said drug manufacturers should have extended investigations to other batches or products and justified it when they classified glass found in product as something other than critical. FDA also faulted them in the areas of sampling procedures after detecting glass in product, adverse trending of glass particulates and breakage, effectiveness of corrective and preventive actions (CAPAs), root cause analysis and line clearance procedures for removal of broken glass.

The three glass-related warning letters FDA issued in 2001-2005 addressed similar issues, as did the four related 1996-2000 warning letters.

“The reasons or the issues were the same, they’ve just escalated over the last five years,” Bloomfield observed.

“This increase in recalls and warning letters indicates a trend,” Bloomfield said. “They’re having a very negative impact on product supply in our domestic and international markets. Our vaccines, they’re really critical, medically necessary products. Our patients aren’t getting what they should get. The cost to industry and the patients is unacceptable. And the shortages of those lifesaving drugs are unacceptable.”

A significant proportion of survey respondents have struggled with failures of glass lots and drug product lots due to glass defects.

Reality diverges from expectations

Amgen’s Martin VanTrieste described results of an industry survey conducted in advance of the meeting that indicated some problem areas.

The survey included ratings of specific glass suppliers, but the data was blinded “behind the scene,” VanTrieste said, so no one can tell which ratings go with which suppliers.

The survey’s 118 respondents were broadly representative of the industry, with 62 percent from pharmaceutical manufacturers and 29 percent with biotech firms. Eighty-five percent were with multinational firms, even though 83 percent were based in North America. There was a range of small, medium and large firms, whether ranked by sales or units produced. Of the respondents, 39 percent worked in quality, 17 percent in manufacturing and 15 percent in research and development.

“Something has changed,” VanTrieste said. “What has changed? Is it glass manufacturing? Is it the source of the cane? Is it the pharmaceutical companies doing something they shouldn’t be doing to glass? Is it distribution? What has changed?”

Certainly, one change is that expectations are higher than in the past, he said. “That’s a message that came through in the survey results: Pharmaceutical manufacturers, biotech manufacturers, feel that the quality of glass is not meeting their expectations.” In fact, 44 percent rated the quality of the glass they receive from their suppliers as unacceptable, poor or fair.

How has glass quality changed in the past five years? A 56 percent majority of the respondents said quality has decreased or stayed the same – even though, as VanTrieste put it, “we all should be embracing continuous improvement to make our businesses better.”

The surveyors asked respondents to rank eight glass suppliers on four criteria: the quality of their glass; their ability to determine root cause; their responsiveness to request for investigations; and their overall customer service.

“Two companies scored above the rest in terms of the ratings. They consistently had 30-plus percent exceptional ratings,” VanTrieste said. Also there was a lone standout in terms of unacceptable and poor ratings.

“The reason we did the survey in this manner, and the reason we reported the results in this way is we believe competition will drive improvements, and here’s a way of just creating a little competitive pressure in the marketplace,” VanTrieste explained.

He cautioned pharma companies, however, that “I do not believe all glass problems start in the beginning of the supply chain.”

More than half the respondents (52 percent) said they would not know if their glass supplier changed its material supply chain. “I find that pretty scary,” he said. “People can change a critical component, probably one of the most critical components of a drug product, making significant changes and you wouldn’t know about it.”

What pharma companies are doing

Nearly 90 percent of respondents said their companies test every incoming lot of glass containers. About 40 percent allow for pre-delivery or tailgate samples to arrive prior to the shipment, while two thirds allow these samples to arrive with the shipment. Three quarters pull physical samples from the actual shipment.

Some 63 percent of respondents said their companies use the square root of N plus one rule for their sampling plans.

Asked what percentage of glass receipts failed acceptance criteria for defects over the past three years, 45 percent of respondents said less than 1 percent, while 38 percent said 1 to 5 percent.

“What really surprised me was that 16 percent of respondents said more than 5 percent of their receipts failed their acceptance criteria,” VanTrieste said. “That’s something that has to be worked out between the … suppliers and the producers of the drug product because [it] means that we don’t have agreement on the specs at that point.”

Some 40 percent said they inspect 100 percent of glass after receiving and approving it but before filling it.

“Why is that necessary if you have a good spec, if your supplier’s producing to your spec, if you’re embracing continuous improvement, why are we testing in quality?” he asked. “Because clearly that’s what we’re doing: We’re testing and inspecting quality that’s not built into the system.”

Over the past three years, what percentage of glass lots failed these pre-filling inspections? Nearly half the respondents who conduct them said more than 1 percent. For 12.5 percent of respondents, the failure rate was more than 5 percent. “That’s scary,” VanTrieste said. “What about the other 60 percent who aren’t inspecting their glass? I would imagine the defect rates are probably pretty similar if you’re inspecting or you’re not.”

The glass-related quality defects have forced drug makers to reject a surprisingly high percentage of product lots over the past three years as well, the survey showed. “You can still see, even with all that inspection going on, 35 percent, 36 percent of the respondents felt that over 1 percent of their drug product lots failed a spec because of the glass defects. That’s a high number. Anytime you see greater than 1 percent of a validated process failing, you ought to ask yourself what’s going on.”

What about complaint rates for broken or cracked glass? For vials, 32 percent of respondents said they receive complaints regarding one to five of every million vials, while 9 percent reported a complaint rate of 5 to 50 ppm.

For syringes, the complaint rates were slightly worse, with 40 percent reporting 1 to 5 ppm, 7 percent reporting 5 to 50 ppm and 2 percent reporting greater than 50 ppm complaint rates.

One third of respondents said that whenever they receive a single product complaint regarding a critical glass defect, they alert regulatory agencies by issuing an FDA field alert, a biological product deviation report, a medical device report or a European Medicines Agency defective product report. VanTrieste added that 61 percent do not and 6 percent take a risk-based approach to filing such reports.

Slightly more than half said they have defined thresholds for reporting product complaints regarding critical defects.

Participants kept circling back to the question of what to do when someone complains about a single unit with a critical glass defect.

Twenty-eight percent of respondents have conducted at least one recall in the past three years due to glass defects, which VanTrieste figured cost industry more than $100 million.

Given the issues manufacturers are having with glass, are they considering switching to plastic? “Only 30 percent of the industry respondents are considering switching to plastic, and only about 14 percent have an active program in place,” he said. “It’s an interesting dynamic. We have changing expectations, we have glass going through the end-to-end supply chain clearly not meeting specifications or expectations, and it’s a hot topic as I look at 270 people sitting in the audience. The question is, is that going to force people to think about something else? And some people are starting to think about something else.”

Asked if they have quality agreements with their glass suppliers, 24 percent of respondents said they do not, something VanTrieste said he found “somewhat encouraging but somewhat disappointing. … The encouraging part is 75 percent do. But there is a significant portion of respondents that says, ‘we don’t.’ It’s a critical commodity. It’s important to have those quality agreements. It allows people to understand what those expectations are and to negotiate from those expectations.”

Glass-supplier audit frequency is two to three years for the “vast majority,” or 73 percent of respondents, although 18 percent audit annually. And 3 percent do not audit their glass suppliers. “To me, it’s a critical commodity, it’s not meeting our expectations, we don’t have quality agreements. We should definitely be auditing those kinds of suppliers.”

An FDA perspective

The survey VanTrieste presented described “some of the same things that we’re seeing from a compliance perspective,” said David Jaworski, an FDA consumer safety officer and team leader, Domestic Case Management Branch in the Office of Compliance, Division of Manufacturing and Product Quality.

“We’re always wondering why we have a significant number of recalls of products, and yet, from the information we see, even after you’ve gone through inspections before you filled products, looking at after-filling inspections, looking at your processes and trying to minimize defects, and we still have recalls. And it’s kind of interesting to see what is really going on from your perspective.”

It’s important, Jaworski said, for drug makers to ensure their glass supply is of consistent quality, and that it’s clean prior to filling, that the manufacturing process does not cause glass defects. Also important: container integrity, how it might interact with drug products, how the glass is protected throughout the supply chain.

He discussed the key regulatory provisions:

• Section 211.94 of Part 21 of the Code of Federal Regulations, which concerns drug product containers and closures;

• U.S. Pharmacopeia general chapter <1>, injections; and

• FDA’s guidance on sterile drug products produced by aseptic processing.

If a manufacturer is finding defective glass containers or a loss of container closure integrity, “what we’re expecting you to do is conduct an investigation, try to find a corrective action and implement that,” he said. “If you have to go to tightened sampling, you need to go to tightened sampling to ensure that that correction is effective.”

Jaworski called attention to the potential public health impact: Cracked vials could become contaminated or lose their contents prior to injection; containers could interact with product to reduce its efficacy; and glass particulates and delamination could cause cardiovascular damage and provoke immunological reactions.

He gave some recall data, presented by recall event rather than by recalls of each individual product. “You can see over the years, in 2008, you’re still seeing glass and particulates. More recently in 2010 and 2011, seems like there’s an uptick, but pretty consistent overall.”

He noted that “syringes in 2010 were a major portion of the recalls.”

“Also, we’ve had some we haven’t really recalled either. Because they’re medically necessary, we’ve allowed them to stay on the market. … They have different things wrong – shipping and packaging defects (in 2008), and various products in 2011 for particulate matter and delamination.”

He said that common issues with many of the recalls include the observation of visible defects, such as metal particles, damaged containers, silicone, excess silicone in containers, foreign matter and leaking containers.

He said the agency has found that visual inspection processes vary, resulting in inconsistent quality from batch to batch and from company to company. “When we go and do an inspection, we’re finding everything from a rudimentary, ‘oh we just look at it under a light by hand,’ all the way up to electronic inspection of your products. And we see this as something that should be looked at.”

FDA is also finding inadequately dev eloped formulations. “We see drug product precipitate … and glass lamellae formation.”

The rise of the delamination issue

Jaworski showed data on injectable drug product recalls dating back to 2003. The data, pulled from FDA’s Recall Enterprise System database on Jan. 20, shows these recalls were on a downward trend until 2009-2010, when they spiked.

“But if you look at injectable products recalled due to particulates, it seems like it’s a low level just going along pretty steady until just recently, when we had the glass lamellae issue showing up in 2011.”

Prior to the recent surge, glass delamination had triggered only one drug product recall, fluorouracil, in 2005. However, several field reports between 2005 and 2010 reported formation of glass lamellae. “This leads us to conclude [that it’s] not a new problem. It’s been around for a long time. The presence of glass lamellae is underreported. The delamination phenomena for the drug product recalled may not have been adequately studied during process development, and some of the delamination may have been caused by changes in the glass manufacturing processes. And vials from at least three suppliers have been associated with delamination. Firms need to consider better supplier controls.”

Jaworski noted that technical literature on delamination dates back to the 1980s, and some textbooks even discussed delamination. “They didn’t call it that at the time. They called it glass flaking. But it goes back a long time. Most industrial pharmacy books will cover this topic to some extent or another.”

He called attention to FDA’s March 25 advisory alerting drug manufacturers to certain manufacturing conditions that favor formation of glass lamellae in injectable drugs:

• Glass vials made by tubing processes, which are hotter than glass molding processes;

• Drug solutions formulated at high pH and with certain buffers, such as citrate and tartrate;

• Longer storage, as duration of glass/product contact correlates directly to lamellae formation;

• Room temperature storage – “when it’s refrigerated, you usually don’t see it;” and

• Terminal sterilization, which he said, “has a significant effect on the glass stability, probably for injectable products that are sterilized terminally, probably the worst thing you can do to impact lamellae formation.”

The advisory recommends that for products at risk, manufacturers should minimize vial surface alkalinity through selection, qualification and quality control of incoming glass vials.

Jaworski explored how manufacturers can prevent glass lamellae formation. “It starts with effective product development,” he said – identifying critical quality attributes and validating the manufacturing process. “Experiment until you understand the process,” he said. “In many cases, too many people are going with three lots and that’s it. You can’t just go with three lots anymore. They really want you to understand your process. If there’s something that shows up, study it. Don’t just let it sit there. And don’t wait until it goes into manufacturing to resolve an issue just because of time constraints.”

He suggested developing realistic process models and effective in-process controls and testing schemes, as well as effective employee training programs.

Drug makers should audit their glass makers, he stressed. “And you cannot abrogate that to a paper review or no review at all, as we’ve seen from the survey.”

They should make sure the glass makers’ acceptable quality limits for incoming glass lots are adequate. And they should apply their own quality control requirements for incoming glass. Also, they should build quality into their manufacturing operations “to prevent rough handling, which causes in-process glass stresses and damaged containers. Is there something you can do to work on your manufacturing operation to minimize the glass damage that could occur?”

There should also be extra stability program scrutiny for products found to be at risk.

On the question of manufacturability, he stressed that the equipment must be suitable for handling glass containers. They should develop manufacturing processes that put less stress on glass containers. “How many people are putting particulate matter requirements on their filling equipment vendors? How many people are looking at a manufacturing process and trying to establish that limit for that manufacturer of that piece of equipment to prevent glass abrasion that could lead to cracks or damage? Those are things that I think you really have to look at from the manufacturability aspects.”

In the area of detection, he urged manufacturers to develop better cleaning methods to reduce particulates in product, better in-process controls and measurements, better, more standardized visual inspection processes – and automated alternatives to visual inspection.

He highlighted the use of continuous process verification, set forth in FDA’s recent process validation guidance, in anticipating, correcting and preventing GMP problems.

Amgen’s delamination case study

Rob Swift, a veteran from the glass container manufacturing industry who joined Amgen eight years ago, described Amgen’s recent discovery and investigation of a glass delamination problem.

When an unrelated glass breakage issue led Amgen to reinspect several drug product batches more than three months after they’d been filled, the company found some “atypical” particles. They were not the type of shards that occur with breakage, and turned out instead to be glass lamellae from the inside surface of the vials.

During the reinspection, Amgen found lamellae in 0.01 to 0.04 percent of the vials, Swift said. “It was a very small proportion of the vials we actually inspected, but it was not something we could ignore.”

Using camera-based counting and tracking, Amgen identified 100 to 750 lamellae in each vial that had them. The lamellae were about one micron thick, and 15 microns to 1 millimeter in length. Their physical characteristics were unusual, Swift said, showing one that had folded over on itself. “We’re not really sure whether that’s because it’s a flexible material or just because it’s so thin that it can fold over like that.”

X-Ray diffraction confirmed that the lamellae were composed of borosilicate glass.

Amgen examined the inner surfaces of the emptied drug product vials that had lamellae. The side wall of the vial just above the heel showed areas of delamination. This was where materials tended to condense after glass makers vaporized them from the bottom of the vial. These materials slightly changed the chemical composition of the side wall, increasing the susceptibility of the glass to interaction with the formulation.

Swift showed an image of the damaged inner surface of one vial. “We have a boundary area that is pitted above the area where the lamellae had separated, and another area that looks like it’s ready to come off. I call this the dry lake bed effect. It had a hydrated surface, which after we emptied and dried the vial, that hydrated surface dried out and cracked.”

Given that the delamination occurred in different batches, in vials from different manufacturers, “it gave us at least reason to think this might be a larger problem than just a few batches, and so we began to look at reserve samples and materials not distributed of a number of batches,” Swift said.

What did Amgen learn from this broader examination? By comparing fill dates and inspection dates for numerous vials, Amgen was able to see that the problem began to arise much earlier in vials from one supplier than from the other supplier it had been using. “With Supplier Y, we saw onset of delamination as early as 99 days, whereas with Supplier X, the earliest indication we saw were vials that were about 400 days old.”

Furthermore, Amgen found that delamination occurred at higher rates in vials from Supplier Y.

Amgen’s Supplier Quality and Quality Engineering groups obtained from both vial suppliers the surface hydrolytic resistance test results for their empty vials. Amgen fitted that data to normal distribution curves and expressed it as a percent of the European Pharmacopoeia’s limit value for surface resistance.

“Our specification at that time was that the vials had to meet the European Pharmacopoeia requirement,” Swift said. “All of the vials that we’re talking about met Amgen specifications.”

However, the vials with surface alkalinity of 20 percent to 70 or 80 percent of the EP limit that was in Amgen’s specs had a much lower propensity for delamination, regardless of which supplier they came from. The alkalinity of vials from each supplier appeared in bell-shaped curves that overlapped somewhat, with vials from Supplier X averaging some 40 percent lower.

Amgen saw that formulation played a key role as well. “We did not see delamination in these same vial populations when filled with other products that had different buffer systems.”

Amgen’s multidose formulation used much less sodium citrate and had a lower pH compared to its single dose formulations, which “reduced the number of batches [with delamination] about in half.”

Corrective and preventive measures

Amgen recalled batches that had experienced delamination. It left batches on the market that might experience delamination but hadn’t yet, and replaced them as batches produced under new, tighter specifications became available.

Amgen reduced the expiration date of product based on “age” data, and stopped using vials from Supplier Y. The company immediately implemented a tightened specification for empty vial surface alkalinity.

Noting that stability programs aren’t designed necessarily to detect problems that occur in small percentages of the population, Swift said, “We enhanced our stability program to improve our ability to detect this.”

Amgen is working with both suppliers to reduce susceptibility of vials to delamination, and find better ways to evaluate vial susceptibility. This analytical development is ongoing, he said, “and I expect there will be some interesting publications related to that work over the coming months.”

He added that Amgen has applied what it learned to prevent similar problems with products in development. “We have now implemented a more robust screening process for the presence of delamination or risk of delamination with products that are in our pipeline.”

How to get off the water wheel

Dan Haines, a research scientist with glass maker Schott North America Inc., showed a famous M. C. Escher perspective drawing of an aqueduct that starts below a water wheel and ends above it, supplying itself with an endless stream of water.

“You can never get off of this water wheel,” he said, pointing to three main factors that influence delamination, with which he labeled each bend of the aqueduct: drug formulation, glass processing and glass composition.

“When someone who’s not familiar with glass tries to look at this and say, ‘where do we get to the root cause,’ it’s kind of a big wheel. You keep going round and round in circles. So what I’m here to hopefully show you is how to get off this wheel.”

Haines talked about measuring glass elements in vial solutions and measuring their ratios, silicon to aluminum, silicon to boron, silicon to calcium – “each of those ratios tells you something about the mechanism of attack.”

For Type 1 glass, the silicon to aluminum ratio should be about 12, so if it’s significantly lower or higher, “that tells you you have an attack going on.”

The starting point for examining the potential for chemical attack of glass is to obtain from the glass maker information such as the alkalinity value, the hydrolytic value and the glass type, he said, noting that there is a range of Type I glasses.

The next step for a proactive study is to do a three- or four-month study of a drug formulation in vials at accelerated conditions. “Look at the glass surface, look at the solution, and you’ll find out ‘do I have a risk, an elevated risk, of forming particulates.’”

If you’re in a reactive mode, trying to find out why particles are occurring, the study works similarly, though with a different objective: “You want to find out what the root cause is to identify is it really a drug formulation issue or is it really something associated with the manufacturing process.”

Tips on working with glass suppliers

Anthony Perry, Schott’s director of quality, talked about improving relationships between pharmaceutical companies and their glass suppliers.

One suggestion was to avoid glass treatments as band-aid approaches to problems like delamination.

“Don’t mask the situation by having your suppliers put in treatment. What you want to be able to do is get the vial in the way it should have been made, not for us to use so much energy that we blast the glass apart and then correct it by putting things like ammonium sulfate treatment in there.”

On partnering, he said, “One of the things that I think we could do better between the industries is in the design, handling and defect alignment. I still think today we’re doing a very poor job. … When you’re considering a new product, look to us to say, ‘what really is reasonable, given the process window that we have.’ Don’t ask us to make something that really shouldn’t be made on our processes, and then spend the next four or five years complaining that it didn’t work.”

He also encouraged pharmaceutical companies to learn from their glass suppliers about proper glass handling techniques. “We do a pretty good job of handling the glass, and we would share that information with you.”

In the area of defect alignment, Perry stressed that drug makers and their glass suppliers should agree where to draw the line on quality – and stick to it. “What tends to happen is you’ll see the customer will be thinking about one quality here even though we signed the agreement, and then there’s some erosion and changes in perception as to where that quality actually is.”

He added that Schott is trying to facilitate this process by “trying to be a little bit more upfront about the defects we can create, and really assessing the risk attributed with those.”

He gave an example of pressure marks. Schott has developed a document that describes them, notes some of the terms customers often use to describe them, explains how they occur and what risk they pose.

Boris Schmid, corporate QA director, Stevanato Group, explored a broader risk-based approach to glass quality.

He said that the traditional approach focuses on the step-by-step improvements in the process of converting raw materials into glass products and on transportation to the pharmaceutical manufacturer.

“But only if we address all the steps can we really assure patient safety,” he said. That means also addressing raw material, pharmaceutical company and lifecycle issues. He called attention to pharmaceutical manufacturing processes such as filling, washing and depyrogenation. “We all know there are risks, but I’m not here to criticize, I’m here to offer a solution.” He went on to describe his company’s quality model.

In a Q&A session, there was some discussion about why there does not seem to be a problem with delamination in Japan. “They have different view on how glass should be handled … and with respect to inspection within the conversion process … the majority of the information is of course confidential,” one participant said. Others had thought Japan’s additional restrictions only concerned cosmetic defects.

Glass-to-glass contact

VanTrieste stirred the pot, noting that a couple of glass suppliers had told the conference glass-to-glass contact is bad. “But you offer glass-to-glass shipment. Why do you do that if it’s not good?”

Schott’s Perry parried thusly, “We can’t get you to change is more the problem. If we could get you to change, believe me, we would.”

Stevanato’s Schmid brought in a historical perspective. Glass makers used to ship vials in partition cells, he said. But the cardboard partitions left particulates on the vials. They began shipping vials shrink wrapped glass-to-glass in response to pressure from pharmaceutical companies to ship vials that were free of these particulates. They could work on ways to ship with a particulate-free method of separation. “There are some issues we would have to work through,” he said. “But it’s certainly one to take away for as an action plan.”

The dream of unbreakable glass

Mark Paviglianiti, director, supplier quality assurance, Merck Sharp & Dohme, highlighted the incongruence between drug makers’ responsibility to patients and the inherent limitations of their manufacturing and inspectional processes.

Customers don’t want vials that let the product leak out or contamination seep in, or that had to be recalled out of fear that they might, he said.

“Our patients really don’t want to hear of our problems. … They want product where it’s supposed to be, on time, all the time and in the right quantity. It’s our problem to get it there,” he said.

Drug makers inspect vials for defects such as cracks that could allow these problems to occur, he said. Typically they take an acceptable quality level (AQL) approach to these inspections, which involves acknowledging that there will be defects and they won’t catch all of them. “Saying it that way, we know that there are defective units as part of the business model.”

Despite these acknowledged limitations, he said that pharmaceutical companies try to have zero defects going out the door.

“Thinking about product safety, patient safety, it’s really incumbent on us to make sure that we do not knowingly have any cracked vials. We just have to do all that we can to investigate, shut the lines down, and have the corrective actions that prevent any type of cracked vials or any type of container closure integrity,” he said.

Consequently, he said, there is a gap between the reality that some incoming defects will escape detection, and the expectation that pharmaceutical companies will only send defect-free product out the door.

“Then when there’s a customer issue with a cracked vial that we receive, we take that extremely seriously, investigate it extensively, more than likely we’ll report it to the agency,” he said. “So you can see the level of scrutiny that we provide for a single event around a cracked vial.”

Aside from the cost of conducting recalls, which he put at several times the $100 million VanTrieste had estimated, there is the patient risk they imply.

“If there’s a recall situation, that means people have received the product, have been using the product, still are using the product until you detected that problem, made the decision, recalled the product. … So the patients are at risk.”

Why is this important? “If the sterility is impacted, it may lead to an infection in the patient or worse, in the injection site or an entire body system could be affected. It could lead to significant illness or death, and they are taking our medicine to get better or prevent disease, not to get worse. … That’s really a breakdown in the credibility we have as an industry.”

Paviglianiti suggested that, while it is important to continue improving the inspectional process, “it cannot be the main or only focus in the pharmaceutical industry or the glass industry. Together we need other solutions.”

For example, he suggested that the glass and pharmaceutical industries could work to develop “a container that just simply does not break.”

He went on to give six key principles for future directions:

• Establish quality management systems;

• Move away from defects as an acceptable cost of doing business;

• Transition from an acceptable quality limits (AQL) approach to controls that focus on defects per million;

• Instead of accepting a defect rate of one in 1,000 cracked vials, strive for less than one in a million defects impacting container closure integrity;

• Do not rely solely on component inspection systems for glass container quality, but rather build quality into the process and focus on making products right the first time; and

• Be proactive “so when there’s a problem in a pharmaceutical production line, you have already thought through and proactively helped manage that.”

Breaking the chain of flaws

Mads Espersen, sourcing quality specialist with Novo Nordisk in Denmark, said any glass surface has a chain of flaws, and is only as strong as the weakest point, as described by a continuous Weibull or log normal distribution called Weibull modulus.

He added that under the Griffiths equation, the deeper the flaw, the less stress is required to make it grow.

Put together, the two equations say that any handling or processing of glass reduces its strength by adding flaws and by deepening existing flaws. Eventually, the glass may come under a load energetic enough to propagate a flaw, breaking the glass.

Hence glass breakage in a process is a function of prior handling and current load.

“My proposition is that we should deal with glass breakage in a preventive way rather than a reactive way because the glass strength in any step of the process is very much depending on what you’ve done to the glass upstream. The tougher you’ve been on the glass, the weaker the glass and the higher the risk you have of breakage.”

Espersen said glass fracture analysis, or fractology, can sometimes identify the root cause of a glass breakage incident. But it other cases it cannot, and all the detrimental factors must be evaluated and corrected.

When glass breakage goes up, he said it means that a step in the process has become hard on the glass, and glass that didn’t break is also impaired, and more likely to break in subsequent processing steps.

Espersen’s key recommendations were to:

• Avoid glass-to-glass contact, “which is the worst because it excites all the resonance frequencies of glass,” for example by reducing glass-to-glass handling, avoiding dead zones in junctions between conveyors, scrolls and wheels, avoiding misalignments and carefully removing glass debris;

• Avoid impact loads, for example by validating glass transportation, depalletizing very carefully, making sure things are properly aligned in washing and filling lines, and avoiding shifting levels between conveyors and dead plates.

He encouraged people to listen for glass impact loads. “It’s very easy to hear in the production when you impact the glass. You really don’t need a billion dollar computer to find that.”

Zero cracked vials at Bayer

Mike Baldwin, director of quality assurance for final release of product at Bayer Healthcare’s Emeryville, Calif., facility, said Bayer got the message from regulatory agencies that they expected its manufacturing process to produce zero cracked vials.

Through a concerted effort, Bayer was able to reduce cracked vials by 75 percent, and routinely manufacture batches with zero significant glass defects.

After analysis in 2009 of a disturbing glass vial breakage trend, Bayer established a glass monitoring program for the facility, which makes lyophilized drug products in 3cc tubing vials at a rate of 280 vials per minute.

The monitoring program consists of weekly meetings chaired by the quality VP or head of compliance that review the prior week’s inspection results and all glass-related issues – even events that are below deviation thresholds.

An investigation into cracked vials identified the vial washer as the source, so Bayer replaced the vial washer.

There were other incremental improvements as well: daily set-up guides, the use of a thermoplastic material called “PEEK” in areas where glass had been contacting metal, a new tray loader, a new accumulator table and an automated vial vacuum tester.

Inspection and sampling myths dispelled

John Shabushnig, a senior manager and team leader with Pfizer Global Quality Operations, dispelled some myths about inspection processes and lot quality.

He said that contrary to popular belief, 100 percent visual inspection does not detect and eliminate all visible defects. The probability of a person detecting a particle ranges from 100 percent for 200 micron particles to zero percent for 50 micron particles.

Another myth is that the term “acceptable quality level” refers to a maximum percentage of defective units in a batch that a sampling plan with that AQL will accept.

But an AQL of 0.1 does not mean a 0.1 percent defect rate, he said. Rather, it refers to the defect rate where 90 percent of the batches are accepted, which is a measure of falsely rejecting good batches.

There is another term, the “unacceptable quality level,” the defect rate where 90 percent of the batches are rejected, that is a better measure of the outgoing product quality, he said.

The third myth Shabushnig dispelled is that there exists a sampling plan that guarantees the complete absence of particles from a batch. There’s no such thing as AQL of zero percent, he said.

Cracked vials told story

Keith Bailey, director of sterile process technology and engineering at Merck Sharp & Dohme, talked about the difficulty of catching rare glass defects on filling lines.

“Once you have less than a handful of defects per million, you can watch a filling line for a month, more than a month, and maybe never see that vial that’s going to become a customer complaint of cracked and broken vial.”

But damaged vials that customers returned can be very enlightening, he said. “One of our big breakthroughs was the detailed examination of damaged vials returned from customers.” Merck saw that the most common problem was a percussion cone-type of fracture, typically near the heel of the glass, that the company surmised was from glass-to-glass impact, since there was no other material at the fracture site.

Merck also saw that one third of all the complaints “were for multiple damaged vials in a single 10-vial carton, and often we’d find broken and cracked vials together.” The difference between cracked and broken vials was simply the magnitude of the impact.

“Like many of you, we have large accumulator tables, and at every step in the process, our vials are being shuffled. So having multiple in a 10-vial carton kind of tells us it wasn’t coming upstream of the cartoning operation.”

If the problem had been upstream, there should have only been one in a billion cartons with more than one damaged vial.

However, it was not easy for Merck to reach the conclusion that the problem was in packaging or distribution, due to confounding factors. Many years of field testing had never produced a damaged vial. Complaint reports often said nothing had leaked in or damaged the cartons. And there were higher breakage rates for some products and filling lines. “But through all of this discussion and talking, we just thought the statistics of getting that multiple in a 10-vial outweighed everything.”

Merck experimented on ways to improve the robustness of vial-carton combinations using an apparatus the company designed, a pneumatic cylinder that would slam the cartons against a hard surface. “Essentially we could dial in the force that we wanted to slam it with.”

Merck saw damage at impact speeds below 10 mph, which could conceivably occur, for example, with forklift trucks in a warehouse.

Using this approach, Merck introduced several changes to reduce damage to vials. It used thicker labels and placed them lower on the vial, closer to the vulnerable heel area. Merck also began using thicker cartons and thicker circulars, the dividers that separate the two rows of vials in each carton.

Merck also re-emphasized to employees the importance of handling the products carefully and reporting any damage. The company also switched refrigerants from frozen gel packs, which don’t always freeze flat, to foam bricks.

Merck put the changes in place on July 1, 2010. Since then, he said, showing results through March 2011, “we’ve gotten over a 50 percent reduction in our number of cracked and broken vial complaints.”

Complaints dropped furthest in March, and data from April and May show that “we’re continuing to even lower level. So for the past three months from that pre-July 2010 baseline, we’ve had over an 80 percent reduction in cracked and broken vial complaints.”

Bailey said Merck really benefited from its end-to-end approach to the problem. Also, its glass-smashing experiments really helped the company find ways to improve the situation. And Merck continues to look for additional root causes so it can apply additional corrective actions and reduce the problem even further.

Tailgate sampling defended

One issue that came up repeatedly throughout the conference was the pharmaceutical industry’s use of tailgate samples, also called pre-delivery samples.

An attendee who works at a Merck plant said his group does very little tailgate sampling due to guidance from FDA over the past couple of years.

Merck’s Bloomfield added that she heard FDA officials tell another conference that they would no longer accept tailgate samples. “Also, we’ve seen an increase in regulatory observations on tailgate samples,” she said.

A Pfizer attendee argued that there should be a place for tailgate sampling. “I hope we don’t throw out the baby with the bathwater. Tailgate I think is an asset if done properly.”

He explained that the alternative, which is sampling directly from the lot, can cause glass quality problems. “We go to great pains to have suppliers pack these vials where they can be handled without breakage, etc. Once you start going into the lot sampling, you can never secure those loads as well as you receive them from the supplier sometimes. As you handle them throughout your process through your plant, you can cause damage that may not have been there in the beginning.”

Schott’s Anthony Perry defended tailgate sampling as well. “If you don’t use tailgates or a method like that, the sample you take at best is going to be something like square root of N plus 1, which is almost no sample at all. When we do that job for you, we give you samples that truly represent the population that they came from. So that’s one thing. Of course that’s got to be built on trust, because if you don’t trust us to take the samples, then there’s a big issue there. But then if you don’t trust your supplier, get a new supplier.”

One participant noted that FDA’s concerns about tailgate sampling stem from the incident a few years ago when heparin had been adulterated. “It became clear that in certain situations, tailgates for raw materials or other products in general might not represent the actual material, could even be adulterated, for instance.”

He suggested that “there’s a lot of value in a lot of the tailgate programs that we have had or want to go to, especially in sterilized products, especially in contamination issues, where you don’t want to open the product for contamination reasons.” But he said FDA’s “blanket statement” against tailgate sampling has caused companies to abandon the practice.

FDA’s Jaworski acknowledged the concerns. “I understand there is some benefit to possibly having tailgate samples.” He confirmed that the policy against tailgate sampling “was mostly related to the potential for contamination of raw materials in transit and not being held in a qualified state during transit.” He explained that the focus was on APIs or excipients that might have been adulterated or substituted, and therefore had to be sampled directly.

“Now when it comes to glass, in glass in this situation, you have to qualify your vendor and you have to be able to make sure you understand that how you get the material from your vendor is appropriate to even consider a tailgate sample being acceptable. So you have to be able to qualify that process, the transportation, all the way through to ensure that you could possibly go to a tailgate sample. And that’s going to be product by product, component by component.”

Additional outcomes

The conference produced some additional outcomes. VanTrieste encouraged people to volunteer to help update PDA’s Technical Report 43, “Identification and Classification of Nonconformities in Molded and Tubular Glass Containers for Pharmaceutical Manufacturing.”

He also suggested that PDA should start a glass handling task force, which could put together a “points to consider” document.

There were a number of industry activities likely to flow from the meeting as well. For example, VanTrieste said he was “shocked to learn that there’s no validation on the transportation lanes that we ship this very precious material around the world in. I can tell you probably next week, all glass suppliers are going to get 10 phone calls saying, ‘I need to do a transportation validation study.’ That’s part of the outcome of a meeting like this.”

By Bowman Cox