Heparin Adulteration Triggered Pharmaceutical Identity Crisis

Pharmaceutical identity crisis arises after ingredients are mimicked by cheap, deadly substitutes. Chemical sleuths tell how they found the melamine cyanurate, oversulfated chondroitin sulfate and diethylene glycol that were hidden in pet food, heparin and cold remedies. Efforts to better identify drug ingredients takes on new urgency with melamine spreading to infant formula and rumors of possible attempts to devise a new heparin adulterant. As instrumental compendial tests are added to monographs, industry, legislators, regulators and compendial organizations grapple with broader implications of this new type of adulteration. U.S. Pharmacopeia leadership talks about establishing a massive standardized spectral library that could be accessed using remote analyzers to instantly identify ingredients, impurities and adulterants.

A profound identity crisis has rocked the pharma­ceutical industry after someone in China oversulfated cheap chondroitin sulfate and substituted it for real heparin sodium API starting materials.

By law, manufacturers must test every lot of a drug product to confirm the active ingredient's identity before putting it on the U.S. market.

But even though it was present in concentrations as high as 27 percent, the heparin adulterant easily fooled the required compendial tests.

Tragically, FDA found that in the first five months of this year, 146 U.S. patients treated with heparin sodium suffered hypotension or allergies - adverse events the adulterant could have caused - and later died. There were only three such deaths in 2007, as in the year before.

The heparin case raised the specter of a new type of threat facing the globalized pharmaceutical industry, coming as it did in the context of similar incidents involving adulteration of glycerin in over-the-counter medicines with diethylene glycol, and of pet food and infant formula with melamine and related industrial chemicals.

Although laws for ensuring drug quality bristle with measures added in response to a series of crises over the years, they never have addressed cases like these, where it appears suppliers secretly substituted indistinguishable cheaper ingredients for economic gain, regardless of the health consequences.

Victimized by this new form of identity theft, the industry and those who regulate it and set standards for it are struggling to prevent a recurrence.

Notably, they are reviewing standard identity tests in a new light. Most of these tests date from an era when such a fraud would have been unthinkable because operations were concentrated at the end product manufacturer's site rather than scattered among suppliers around the world.

Many involve classic wet chemistry methods, which are considered generally less specific and more subjective than modern instrumental methods such as spectroscopy, chromatography and electrophoresis.

In all three cases - heparin, glycerin and melamine - authorities are buttressing identity testing requirements in the U.S. and internationally.

Meanwhile, industry, legislators, regulators and compendial organizations are all grappling with the broader implications of the crisis. How shall they go about confirming the identity of any other pharmaceutical ingredient that someone might substitute with a cheaper substance that could pass legally required tests?

A confounding factor is that such profit-driven ingredient substitution is likely to focus on the costliest ingredients that can be mimicked on a massive scale by cheap compounds. It is a scenario at odds with typical pharmaceutical testing regimens, which involve tight specifications that are imposed without regard to ingredient costs or market conditions.

However, in the absence of a specific threat there are disincentives to adding tests such as instrumental methods that can "fingerprint" ingredients or likely adulterants.

It could be prohibitively expensive to test for every adulterant of every ingredient that might be introduced by scheming suppliers, even if it were possible to anticipate them all.

Adding more-specific identity tests to new drug applications would be possible, but it would take time. There is always the risk with proposing changes to NDAs or ANDAs that FDA might use the opportunity to address long-simmering concerns about other aspects of the applications. There is also the worry that new tests might reveal previously unidentified impurities that would then have to be studied and perhaps removed. And there is the cost of the instrumentation itself.

Proposing pharmacopoeial monographs for new identity tests and assays would establish them as public standards, enabling other parties throughout the supply chain to better safeguard it from fake ingredients. However, that process takes time, and because of intellectual property concerns, manufacturers are often reluctant to begin pursuing it until their market exclusivity is expiring.

For now, some manufacturers say privately that they are simply adding more instrumental testing, even though it is not legally required.

In any case, the key for manufacturers, Moheb Nasr, director of FDA's Office of New Drug Quality Assessment, told "The Gold Sheet," is to update their identity tests for drugs and biologicals, making sure they include a highly specific characterization test as well as a functional bioassay test, although the latter may not be needed for small-molecule drugs.

The extent of the industry's response is unclear. When Amgen's Martin VanTrieste asked for a show of hands in September at the PDA/FDA Joint Regulatory Conference in Washington, very few participants in a corrective and preventive action (CAPA) session indicated they were using the CAPA process to prevent a heparin-like incident with any of their products.

Yet sessions on the heparin crisis have been drawing crowds, including one at the U.S. Pharmacopeia's Sept. 23-26 annual scientific meeting in Kansas City, Mo.

At that meeting, USP unveiled an initiative to enable widespread screening for counterfeit ingredients or drug products, even in poor countries with little in the way of sophisticated laboratory testing capabilities.

USP officials talked about creating a massive spectral library that manufacturers and regulators could access via handheld remote scanning instruments to instantly identify a plethora of ingredients, impurities and adulterants.

However, there is some concern that such an approach could wind up being too specific. Many ingredients can be made different ways, and so can have a variety of spectra. If the library doesn't include every type, it could brand some legitimate ingredients as counterfeit.

Meanwhile, Congress appears poised to push through legislation next year that would impose new requirements on FDA, USP and industry to confirm the identity of drug ingredients.

How P&G Identified Melamine in Pet Food

Roy Dobson of Proctor & Gamble's Analytical Global Capability Organization in Mason, Ohio, riveted USP conference attendees with his account of how P&G solved last year's pet food adulteration mystery.

In that case, perpetrators fooled the standard Kjeldahl method for assaying the protein content of food. The method, which Danish beer chemist Johan Kjeldahl introduced in 1883, actually measures nitrogen, not protein. Even though protein is rich in nitrogen, other substances are as well - and not all of them are benign - as Dobson and his colleagues at Proctor & Gamble found.

It took the coordinated effort of several P&G labs, each with its own distinctive capabilities, to solve the mystery.

The first breakthrough came the night of Thursday, March 22, after a futile search of contaminated wheat gluten for certain mycotoxins associated with wheat products that are potent nephrotoxins even at part-per-billion levels.

P&G was examining aqueous extracts using hydrophobic interaction chromatography (HILIC) in conjunction with an ultraviolet detector in series with a quadrupole mass spectrometry instrument.

This approach enabled the researchers to observe some unusual polar species lurking in the void volume that would have escaped their notice with conventional reversed-phase liquid chromatography, Dobson explained.

"We were able to flag a few suspect ions that seemed to be associated with this particular peak,' Dobson said, "and we literally - and I'm not kidding - ran upstairs to where we were using the time-of-flight instruments to do some general broad-based screening for unknowns."

They halted a reverse-phase HPLC screening run that was under way and looked in the void volume, where "we were able to find a (calculated mass of) 127 ion that correlated to that peak that we saw in the quadrupole experiment," Dobson said.

Some quick data crunching showed there were four possible elemental compositions, of which they could deductively rule out all but one: C3H7N4.

"That can't be right," he recalled one of the other researchers saying. "You can't have that many nitrogens and double bond equivalents in a molecule that small."

He flippantly replied, "Oh sure you can," and sketched a possible molecular structure. The others raised their eyebrows. "Well, let's check," he said, and called downstairs to an associate who had been doing some literature searches in SciFinder, the American Chemical Society software for searching its Chemical Abstracts Service database.

When he plugged in the molecule Dobson had proposed, "Bingo," he told the USP meeting. "There was the same structure I had drawn just a few minutes before. It had 10,000 hits in the literature." It was melamine.

Triazines Presented a Lingering Puzzle

Instead of solving the mystery, however, they had only deepened it. And as they completed additional confirmatory testing, it grew deeper still.

The technical literature showed that melamine was rather benign and easily eliminated due to its high solubility in water. So how could it be wreaking such havoc in the kidneys of pets?

A key hint came when the P&G researchers reconfigured the time-of-flight instrument for the higher resolution HILIC chromatography and it showed that other triazine-related compounds - not just melamine - were present.

Dobson is certain that if they had just used a more conventional lower-resolution instrument with a nominal mass or single mass unit display, the whole family of triazine compounds in the tainted cat food "would have been buried in the complex matrix ions." These crucial clues might have been completely missed.

But the presence of the other triazines - ammeline, ammelide and cyanuric acid - did not immediately suggest a mechanism of toxicity.

Melamine and cyanuric acid "are commodity chemicals, they're seen as relatively safe, they're almost ubiquitous out there in various applications," Dobson said. "Triazines themselves, as we searched the literature, separately have been shown to have relatively low toxicity, nothing like what we were seeing here."

Still, there were lingering questions among veterinary medicine thought leaders, P&G toxicity experts and FDA, Dobson said. Were they missing something? Could the triazines be markers of some other chemical? Or were they uniquely toxic to cats?

To find answers, they did more testing.

The P&G team started with a single cloudy 400-microliter urine sample collected by a veterinarian from a cat 10 days after it had stopped eating contaminated food.

Because the literature indicated that the triazines would all clear within 24 hours, "We didn't have a lot of hope that we were going to be able to detect anything of importance," Dobson said.

But when P&G ran the cloudy urine through a centrifuge, re-suspended the insoluble particulates in water, and assayed it for the four key triazines, it found they totaled some 0.2 milligrams per milliliter. "This was an astonishingly high level of material."

P&G turned to its Woods Corners Technical Center in Norwich, N.Y., with its solid state analytical capabilities, to distinguish between suspect and control wheat gluten.

Woods Corners saw crystals in the suspect wheat gluten, which it examined using Fourier transform infrared spectroscopy, Fourier transform Raman spectroscopy, nuclear magnetic resonance spectroscopy and X-Ray diffraction. Surprisingly, the crystal spectra did not match pure melamine or any other triazine.

P&G's Miami Valley Innovation Center in Ross, Ohio, used its experience with breast cancer diagnostics to micro-image damaged feline kidney tissue a private veterinarian had supplied. Using Fourier transform infrared spectral imaging, the lab saw crystalline materials that differed from ordinary kidney stones.

Woods Corners then imaged the kidney tissue and determined that the spectrum of the unknown kidney crystal matched that of the gluten particle it had been studying.

The lab examined a number of crystals in the kidney tissue and found they were made of the same material, which was not pure melamine.

The researchers removed some 20 nanograms of crystal from the kidney tissue for HILIC mass spectrometry analysis, which showed a mix of melamine, ammeline and cyanuric acid.

Hypothesizing that the triazines might be combining with naturally occurring counterions to form an insoluble salt, the researchers performed screening tests with acids that are ubiquitous in biological systems, such as uric acid and acetic acid.

Their X-Ray diffraction and FT-IR data found a match with the gluten isolate: it was a crystal formed by two of the triazines - melamine and cyanuric acid. The melamine cyanurate also matched the crystals recovered from cat urine and kidney tissue.

A HILIC-based semi-quantitative assay showed extremely high levels of the triazines in tainted wheat gluten of about 8.4 percent melamine, 5.3 percent cyanuric acid, 2.3 percent ammelide and 1.7 percent ammeline.

Similarly, the assay showed total triazine levels as high as 5.0 percent in replicate cat kidney specimens.

P&G fed rats simulated tainted feed, which confirmed its theory that the bases, melamine and ammeline, were reacting with the acids, cyanuric acid and ammelide, to form insoluble crystals in kidney tubules, where it blocked the flow of urine.

Water solubility was a key to the toxicity puzzle, P&G found. Some 3,240 milligrams of melamine could be dissolved in a liter of water, as could 2,000 milligrams of cyanuric acid. But the same volume of water could only hold 2 milligrams of melamine cyanurate.

In fact, the crystal formed such a highly stable, insoluble lattice that it was patented as a fire retardant.

Amino acid analysis showed how much protein the tainted wheat gluten really contained.

Using a combustion method that originated in the early 19 century and is now fully automated, P&G ran control wheat gluten and tainted wheat gluten through a CHN analyzer to see if they met the 70 percent to 75 percent protein content specification.

The CHN analyzer, which like the Kjeldahl method measures total nitrogen, indicated the tainted wheat gluten as 83 percent protein, compared to 77 percent for the control sample.

However, an amino acid assay, which measured protein directly, produced a very different result. The control sample was 87 percent protein while the tainted wheat gluten was only 39 percent protein. The 18 percent of triazines present almost exactly accounted for the difference.

It is possible, Dobson speculated in response to a question, that some suppliers may already have been beefing up apparent protein content with industrial melamine, and that it was only when someone "got greedy" and began using a cheaper waste product containing the other triazines that the injuries and deaths occurred and the fraud was unmasked.

Noting that there are rumors that this type of adulteration has been going on for a long time, Dobson said, "this is totally speculation, but I think it is informed speculation: It could be that pure melamine has been put in products for a long time and just not been caught."

Perhaps it had been used historically to bump up apparent protein content by a few percent. But in the pet food case, with protein at half the level it should have been, "we're talking about gross addition, so that costs money. So it's conceivable that these folks have gone after scrap industrial byproduct, non-pure melamine, and that's where a lot of this problem has started to surface right now, people are getting greedy."

Search for Heparin Adulterant was Complicated

The heparin crisis early this year precipitated a massive effort to identify the deadly adulterant that had escaped detection by standard compendial methods such as the sodium flame test (see "The Gold Sheet," February and March 2008).

The original focus was on Baxter International, Deerfield, Ill., and Changzhou SPL, its heparin API supplier in China, although it turned out that the problem also affected other manufacturers and suppliers, though to a lesser degree.

The work was complicated by the nature of heparin, a "spaghetti" of molecular chains that is "one of the most complex biological compounds in nature," Ali Al-Hakim, a branch chief and heparin expert at the Office of New Drug Quality Assessment in FDA's Center for Drug Evaluation and Research, told the USP conference.

It is mostly a polymer of two-ring sugars, or disaccharides, each containing a uronic acid and glucosamide. Each can have any of some 32 different combinations of stereoisomers, sugars and sulfation patterns. The molecule, which depends on three dimensional conformation for its bioactivity, can have a molecular weight ranging from 2,000 to 50,000 Daltons.

Ultimately, more than 100 researchers at numerous institutions were involved in solving this deadly pharmaceutical science mystery, most of whom coauthored articles that appeared in late April in The New England Journal of Medicine and Nature Biotechnology.

A number of federal laboratories took part in the testing - the Centers for Disease Control and Prevention in Atlanta, FDA's Forensic Chemistry Center in Cincinnati, and FDA' Office of Testing and Research in St. Louis. Private sector and academic research labs were involved too.

As with the melamine case, labs screened samples to cover a wide range of possibilities. They looked, for example, for porcine viruses, bovine tissues, toxins, peptides and proteins, nucleic acids, histamines, small molecules, solvents and heavy metals.

Techniques used included gas chromatography, high pressure liquid chromatography, enzyme-linked immunosorbent (ELISA) assays, polymerase chain reaction testing, nuclear magnetic resonance imaging, nitrous acid degradation, enzymatic and mono­saccharide analysis, precipitation and electrophoresis.

"We just had this hint at that time," Al-Hakim said. "There is something similar to heparin, it's no small molecule, it's a big one, sulfated, but we don't know what this material is."

Proton NMR and capillary electrophoresis showed peaks that shouldn't be in heparin, he said. Focusing on the NMR spectrum of the heparin molecules' anomeric acetyl region, they saw an extra peak at 2.18 ppm. Heparin peaks at 2.04 ppm and a natural impurity, dermatan sulfate, peaks at 2.08.

Based on those findings, the agency published NMR and CE screening methods, even though they were not yet validated, so that healthcare facilities in the U.S. and around the world would have a way to reject batches containing the as-yet unidentified impurities.

Deadly Ruse Involved Highly Educated Trickery

As the difficulty of identifying the adulterant became apparent, Al-Hakim said it was clear that "the people who were behind this were very clever and highly educated."

The USP test methods, which addressed potency, anticoagulant activity and impurities such as solvents, were not designed to discover such a deviously contrived adulterant.

"The contaminated material actually seemed like heparin," he said. "It was a polysulfated poly­saccharide, so it would pass any test. Regardless of whether it is the NDA test or the USP test, it will pass."

The oversulfated chondroitin sulfate has the same molecular weight as heparin, it is just as sulfated, and it has a similar sugar pattern. "Everything is the same," he said, "so it was very difficult for us to detect it."

But there were clues. While heparinase I, II and III enzymes will normally digest heparin, they were not fully digesting the contaminated samples.

Also, carbon-13 NMR showed an O-substituted N-acetylgalactosamine even though, Al-Hakim noted, "heparin does not have galactosamine." Carbon-13 NMR also showed a signal that implied a glycoside linkage between monosaccharides.

Two-dimensional NMR spectral plots showed that both sugars in the disaccharide unit contained two sulfate groups, something never before seen with heparin.

Without knowing yet what the heparin-like molecule was, Al-Hakim said, "we managed to actually purify that compound and isolated it."

FDA subjected the isolated contaminant to a battery of tests, and used a combination of techniques to map its structure. This led to the discovery of the molecule made up of linked sulfoglucuronic acid and sulfo-N-acetylgalactosamine sugars, which they called oversulfated chondroitin sulfate.

There are still a lot of unknowns. For example, it is unclear what starting materials were used: the cartilage needed can be obtained from fish, sharks, cows, pigs or chicken. Also there remains uncertainty about the reagents used. "Sometimes we found pyridine, which means they did the sulfation with pyridine as a solvent, and sometimes we found DMSO, dimethyl sulfoxide, or dimethyl formalide."

Al-Hakim speculated that someone may have gotten the idea of oversulfating chondroitin sulfate from a paper published in the late 1990s that indicated it would cause the naturally available material to take on some anticoagulant activity.

It appears the adulterant was mixed with crude heparin starting in middle to late 2006, starting at 1 percent and slowly increasing, eventually to as high as 27 percent. The higher concentrations were associated with the most serious side effects in patients.

As heparin prices increased after a swine virus decimated China's pig population, "they started to add more because this was very cheap compared to heparin," he said. "A kilogram of chondroitin sulfate, unsulfated, is $20. A kilogram of heparin is $3,000. So you can see if you even put in 10 percent or 5 percent, you are making thousands of dollars."

It turns out, Al-Hakim said, that a German company had once marketed a similar compound as an arthritis medicine called Arteparon, but then withdrew it due to adverse reactions.

Earlier studies had not shown any toxicity in rats. However, in new animal studies using pigs, a team of researchers from FDA, Harvard, the Massachusetts Institute of Technology and the Rensselaer Polytechnic Institute confirmed a direct link between the compound and the toxic effects.

Although with surveillance using the NMR and CE screening methods "everything went back to normal," FDA is working to do more.

Protection from New Threats Sought

"We are trying to be proactive because we heard that somebody somewhere, whether in China or somewhere else, may be trying to put in another contaminant similar to heparin and similar to chondroitin sulfate, which can pass these tests," Al-Hakim said.

"There are so many polysaccharides out there, from bacteria, from animals, and you can synthesize them, add sulfate and you can make them look like heparin," he said. So FDA is working to establish a test that would catch any heparin mimic that might come along.

Bill Koch, USP's chief metrology officer, told the meeting that FDA had asked USP in March 2008 to work on updating its heparin monograph. "We launched an intensive effort to look at the methods that FDA had posted, at their request, and to validate those methods. So in about three months, USP was able to do a revision bulletin, and we added the NMR and CE tests to the monograph for heparin with the appropriate reference standards available."

But USP knew that was just Phase 1. "That was a stopgap measure to get oversulfated chondroitin off the market," Koch said. "While that was being done, we also continued the effort to better identify heparin, to uniquely define the heparin and the impurities that might be there."

The pharmacopoeia's advisory panel "is now finalizing our Phase 2 approach to look for new tests for the identification and impurities, and the development of a new potency assay procedure that should better define, describe the properties of heparin that are important for heparin," he said.

USP hopes to publish an interim revision in PF 35, with a proposed official date for these new methods in August 2009, Koch added.

DEG Resurfaces in Glycerin

The problem of diethylene glycol-contaminated pharmaceuticals has surfaced repeatedly around the world since 1937, when a Bristol, Tenn., manu­facturer mixed DEG with sulfanilamide and distributed the liquid antibiotic without any safety testing, causing 107 deaths and leading to passage the following year of the Food, Drug & Cosmetics Act.

Subsequent incidents occurred with drugs that contained DEG-contaminated glycerin. "This is something that has happened over and over again," said Albinus D'Sa, a senior compliance officer with the Division of Manufacturing and Product Quality in FDA's Center for Drug Evaluation and Research.

FDA responded to the latest incident, which occurred in 2006 with DEG-tainted glycerin made in China and used in Panama, by issuing a guidance document and launching a task force that came up with some proposals. Meanwhile, President Bush launched an import initiative and the Health and Human Services Department took action as well (see "The Gold Sheet," August 2007, p. 8, and September 2007, p. 8, p. 12).

Noting that the FDA guidance requires manufacturers to test every lot and every container, he explained, "We really wanted methods that were easily available. USP already had a method, and we felt that if additional methods are provided, that would be really nice."

Stressing that DEG is a contaminant rather than an impurity, D'Sa said it has been difficult to pinpoint the cause of the problem and a sure way to reduce the risk to consumers. "Is ingredient testing the solution, or should we test products that have glycerin in them? Are other ingredients impacted? Is this a supply chain management problem? These are questions that we within the FDA have been struggling with."

The Role of Dual-Use Chemicals in China

D'Sa called attention to July 18, 2007, congressional testimony by FDA Deputy Commissioner Murray Lumpkin stressing the agency's understanding "that China does not require registration of chemicals that may have a dual use in both industrial chemical products and drug products."

Lumpkin explained to the Senate Commerce, Science and Transportation Committee that without such registration, "these products are exported as chemicals but then used as starting products for the production of pharmaceuticals."

Because of this, D'Sa suggested, labeling mix-ups could be behind the problem of DEG in glycerin. Consequently, the solution to DEG-tainted glycerin may involve more than just finding a test method.

"There are two other types of products labeled as glycerin," D'Sa noted. "One is industrial glycerin, which contains sorbitol, diethylene glycol and glycerin, in the ratio of 18:4:2.5." This product is sold to printers, he said. "It's called printing ink glycerin, and it has the same amount, it works out to about 18-20 percent, of diethylene glycol, [as] we found in Haiti and in Panama."

Another type of glycerin, called textile glycerin, contains glucose, DEG and water in the ratio of 6:3:1, but no glycerin. "Can you imagine that this is also likely to be a labeling problem too?" D'Sa queried.

"My question then is do we require a distinct label for this? Should we distinguish this by color, cautionary symbol? And assign to test before use? That's my question to USP."

A post-Haiti workshop in July 1997 led to three key suggestions:

· international standards for labeling of APIs, excipients and poisons to be agreed upon and uniformly applied;

· improved standards for the certificate of analysis; and

· new analytical tests that are portable, inexpensive, easy to use and specific and sensitive.

D'Sa said FDA's 2007 guidance essentially concluded "that you need to have a good supply chain, you need to have good, reliable vendors, you need to know who you're buying it from, you need to test, and you need to test every container because the nature of this problem is that good material is mixed with bad material."

During inspections in Haiti and Panama, he said FDA found that when examining material coming into warehouses, "you really can't tell where this bad material is, unless you test every can of glycerin."

Tighter, More Restrictive DEG Limit Proposed

In response to the Haiti incident, USP in the late 1990s added a glycerin identity test (Identity Test B, glycerin ID by retention time) and a limit test for DEG and related compounds to its glycerin monograph, USP's Bob Lafaver told the meeting.

While drafting its guidance on DEG in glycerin, FDA in April 2007 wrote to USP, requesting it to clarify that the DEG limit test is part of the identity test, not a separate impurity test.

The rationale FDA set forth in the letter was that if it were part of the identity test, then cGMP regulations at 211.84(d)(1) would require manufacturers to detect and quantify any DEG present.

"One of our challenges was that there are complex issues related to requiring that contaminants or adulterants be considered part of an article's identification - something that has never been done before in the USP," Lafaver said.

D'Sa noted that in its guidance, FDA recommended using any test method with a detection limit of less than or equal to 0.1 percent. FDA agreed with this limit, which originated from a meeting of the Pan American Health Organization, "because we knew the capability of several methods that would do that."

The guidance recommends using the USP gas chromatography method, but also has a thin-layer chromatography (TLC) method, even though it may not be as robust as the USP method.

USP has worked to verify the TLC method, ultimately concluding it was not robust enough. Although it yielded useful data in FDA's analysis of DEG in toothpaste, D'Sa said, "This is a method that takes some skill and requires you to really monitor."

When USP revised the monograph to make the DEG test part of the identity test, it also proposed a 0.025 percent quantitation limit.

However, USP postponed publication after industry stakeholders said they wanted to do a collaborative study to confirm sufficient accuracy, precision and specificity for that quantitation limit.

The study would see if typical laboratories could meet the tougher standard in demonstrating the absence of DEG and ethylene glycol from glycerin using either gas/liquid chromatography with flame ionization detection or an equivalent method.

Lafaver presented the results of the resulting method performance study in Sept. 18 and 19 USP webinars.

Two overlapping groups of eight labs that met system suitability requirements and followed protocols performed the tests.

Lafaver said results were accurate enough for a non-quantitative limit test, though not for a quantitative test. Precision and specificity were good.

There were no false positives: little or no DEG or ethylene glycol was detected in neat glycerin samples. Also, samples spiked at half the 0.025 percent limit gave readings below that threshold.

Debate Over Proper DEG Detection Limit Resolved

There was a vigorous discussion at the USP meeting in Kansas City about the need for speed and accuracy and the appropriate limit for the DEG testing. Several attendees questioned the 0.025 limit that USP had proposed.

"The real question is: what's the goal?" said David Schoneker, chair of the International Pharmaceutical Excipients Council - Americas. "If the goal is a rapid screening technique … to test every container, isn't 0.1 percent good enough?"

The 0.25 percent limit would be too difficult for many rapid screening methods to meet, he said. While the USP method "certainly has been demonstrated to be robust at that level, [it] is not exactly a rapid screening technique."

Another participant suggested that if the limit is tightened now, it could prevent or stall development of an improved version of the TLC test, one involving the use of concentrated zone plates or some other concentration technique, that could meet the 0.1 percent standard. With such a method, "every drum could be tested on the fly without having to wait for results from the GC analysis."

One participant said her company has a near infrared method that can quickly check for DEG levels down to a 0.05 percent limit and asked whether USP would consider it for an alternative method.

USP Chief Science Officer Darrell Abernethy replied that the pharmacopoeia's expert committee would consider it, but it would need to see the validation data.

Another USP official cautioned that it can be difficult to transfer near IR methods from lab to lab, which can make them problematic for compendia.

USP's Excipients Monographs 1 Expert Committee voted Nov. 19 to change the DEG limit to the 0.1 percent level that FDA's guidance recommended.

How to Define the Needle and the Haystack?

Contamination of the increasingly global pharmaceutical supply chain, whether intentional or inadvertent, has drug makers pondering some big questions, USP's Bill Koch noted at the Kansas City conference.

"How do you find something you're not looking for? When do you start looking for what you're not looking for? Where do you look? If you do find something, do you keep looking for more things, for other things, in other things?"

The chief metrology officer went on to draw a parallel between finding a needle in a haystack and finding oversulfated chondroitin sulfate in heparin, melamine in cat food or DEG in glycerin.

A magnet might remove needles from a haystack, but would be useless against brass screws subsequently introduced. "Now we're chasing contaminants, having to find new ways of finding them and new ways of keeping them out."

Another approach would be to avoid contaminants by identifying and keeping only the hay. "It's easy to identify hay versus a needle. But is it so easy to identify hay vs. grass or straw? They're very similar. Now we need new techniques."

There could also be problems with matrix effects. A technology that could pull needles from loosely piled haystacks might not work with hay that is tightly rolled.

Then there is the question of where in the supply chain to look for problems - before, during or after processing ingredients, or when finished products are in transit or at the point of consumption.

The recent contamination incidents also raise new questions about the roles and responsibilities of USP, drug manufacturers, FDA, police and the FBI, he added.

USP aims, he said, to guarantee the identity, strength, quality and purity of drugs with "scientifically based measurement procedures and methods, our monographs and general chapters that uniquely define and describe the API or drug product to the exclusion of everything else." However, in today's environment, he stressed that this is easier said than done.

Spectral Library Solution Proposed

Likening the effort to chase down every contaminant to an arms race that USP is unlikely to win, Koch suggested a different approach.

"We need to develop a living spectral library of both known and suspected impurities, contaminants and adulterants, so that every time we do find a new kind of needle in there, we need to document it, put it in the spectral library, and perhaps keep it from being used again," he said.

"This is an endeavor that should be incorporated into the USP," said David Bugay, who serves on USP's expert committee for general chapters and is president of a consulting firm, PharmAnalysis.

Speaking in a Sept. 23 session focused on longer-term initiatives that could lead to resolutions at USP's 2010 convention, the spectroscopy expert said why he thought USP should establish a spectral library.

A spectral library is a searchable database of spectral information that can be used to identify collected spectra from a particular material, Bugay said. "Now, this material can be a solid, it can be a liquid, it can be a gas, for that matter. It can be an API, it can be a drug product, it can be a contaminant."

The databases can include infrared and Raman spectroscopy, which Bugay noted "are vibrational techniques measuring vibrational motion associated with a molecule, and give a fingerprint to chemically identify the material."

They can also incorporate data from other techniques such as mass spectrometry, NMR and X-Ray powder diffraction. "Each one of these spectroscopic techniques works with a different physical phenomena, probes your material in a slightly different way."

This approach can enable a flexible, multidisci­plinary approach to characterization: "you can link these various libraries and so you can utilize one technique as a higher priority over another in your identification of unknown materials."

The databases can incorporate textual information such as melting point and usage, as well as scanning electron microscopy or optical microscopy images.

Pharmaceutical companies already are using spectral libraries to identify APIs, excipients and drug products, as well as for stability testing by checking for the unique IR and NMR spectra of degradants.

Potential uses could include identifying drug products and ensuring their stability in remote settings around the world.

But Bugay is most excited by - and most focused on - the anti-counterfeiting potential of spectral libraries, as well as their ability to support reverse engineering of products to defend intellectual property.

One possible benefit to USP of establishing an authentic, standardized cGMP spectral library would be the opportunity to sell access to end users.

Spectral Searching Demonstrated

Bugay noted that there already are spectral libraries made with techniques that have been described in USP monographs - infrared, near-infrared and Raman spectroscopy, mass and NMR spectrometry, ultraviolet and visible-light spectrophotometry, and X-Ray powder diffraction. They have been used to generate libraries of excipients, narcotics, explosives, oils, adhesives, organics and inorganics.

Even though it is a well-recognized, widely practiced area of spectroscopy, it is not standardized, he said, and questions remain about the authenticity of the libraries.

Bugay went on to demonstrate the use of a library of some 1,500 API Raman spectra to identify an unknown API. With its well-resolved peaks and flat baseline, Raman spectroscopy is "an ideal technique for spectral searching." The search quickly showed the top five matches, and the first, carbamazepine, was correct.

He then searched a library of some 300 excipients for matches to an infrared spectrum of a tablet. While it doesn't give as good a spectral resolution as the Raman spectrum, "there is a significant amount of information in this infrared spectrum that does allow us to identify the various components within this pharmaceutical tablet." It showed the presence of cellulose, even indicating the likely grade, as well as cross carmellose sodium, corn starch and gum.

Current Practice is Widespread but Scattered

Although spectral libraries are widely used, the current practice cannot support the type of industry-wide anti-counterfeiting applications that Bugay envisions.

Many existing libraries are specific to a particular platform or instrument, such as Bruker or Thermo Nicolet instruments.

Also some of the libraries are very limited in scope, addressing only organics or oil or adhesives, and often with just one spectral technique.

Bugay also expressed concern about the reliability of many existing spectral libraries. "You don't know how they were built. You don't know where the compounds originated from."

It could be a big problem if the materials used to generate the spectra contained impurities. "These techniques have the ability not only to generate a fingerprint for the major component, but also the minor components," he said.

"Many of these techniques have detection limits on the order of 1 percent, 0.5 percent, 0.2 percent by weight, and so impurities that are in those materials are going to show up in that spectrum and those are not a physical property of the material of interest. It's an impurity, which can lead you astray in your spectral searching. So this is obviously a major point of concern in some of the libraries that are out there."

Also there could be questions about spectrometer configuration, particularly considering existing libraries probably were not built under cGMP conditions, he said.

Bugay estimated that pharmaceutical companies each probably have three to 10 of their own proprietary spectral libraries of APIs and excipients for use in protecting their intellectual property and combating counterfeits.

Remote Analyzers Can Unlock Potential

The advent of remote analyzers is revolutionizing the use of spectral libraries, Bugay said.

The pharmaceutical industry has begun using handheld Raman spectrometers that "are getting the exact same, if not better, performance characteristics as with a research-based instrument, vintage of about five years ago."

However, he cautioned, the instruments are only as good as the spectral libraries that support them - a limitation USP can help address.

Bugay envisions spectral libraries built for materials management for remote analysis of incoming raw materials. "An operator can sit on a loading dock. A 55-gallon drum of magnesium stearate can come in. Don't take it off that tractor trailer until I analyze it here, and then I can release it right here and now as opposed to putting it into a quarantine situation."

Another application is to confirm the source of excipients. "Believe it or not, you can look at isotope ratios in mass spec. You can tell whether the cellulose is from Norway or China or South America. That is valuable information when it comes to anti-counterfeit measures and tracking down the source of these materials."

Yet another application is to confirm the stability of products throughout the distribution chain. Healthcare providers in remote areas could use Raman spectrometers loaded with spectral libraries to confirm potency before administering products that may or may not have been properly packed with dry ice while backpacked into a harsh, undeveloped country.

Over the next two to five years, Bugay expects to see more standardized spectral libraries that can be linked together for multidisciplinary analysis. "In my opinion there's no one single technique that can tell you absolutely unequivocally that you have counterfeit material, yet if you take a multidisciplinary approach, I can tell you unequivocally whether it's counterfeit or not."

He envisions chemometric packages that link together the various databases and emphasize the spectral techniques that provide the most telling analyses of given materials. Also, the spectral libraries the packages access will become linked among multiple sources worldwide.

To use spectral libraries in this fashion, they would have to meet some type of standards, he noted.

The materials used for generating the libraries would have to be authentic and of high purity, the instruments used would have to be properly calibrated and the data acquisition parameters would have to be consistent. There would have to be proper cGMP documentation and the search algorithms would have to be consistent.

Bugay urged USP to issue a general position paper, guidance or general chapter on the use of spectral libraries for chemical identification. It also could promote the use of spectral libraries and update its monographs. "Virtually every API monograph has a spectroscopic and chemical identification technique," he noted. "Let's update the monographs such that we can utilize spectral libraries in the identification component of the monograph testing."

He also suggested that USP commission and generate cGMP spectral libraries, perhaps for specific classes of drugs.

USP could support its NF excipient monographs with a central database of excipient signatures for use in anti-counterfeiting. There could also be a linked database of spectral signatures of the adhesives, packaging, inks and glues used in packaging.

USP could play a key role by providing authentic material for use in building a spectral libary.

The pharmacopoeia could start by issuing a position paper on the use of spectral libraries, then commissioning the building of the libraries.

"I feel very strongly about further utilizing the technique under a more guided path that USP I believe can provide to industry."

Asked about potential application to dietary supplements, which face new U.S. identity testing requirements, Bugay cautioned that spectral signatures of complex materials like botanicals and biologics are a lot more complicated.

Even so, he said that with a multidisciplinary approach that links libraries and chemometrics, "I feel very strongly that even complex molecules can be identified with this approach."

In fact, he said, "I feel that a combination approach here would be available to identify follow-on biologics vs. the innovator product."

There may, however, be other hurdles for using a public spectral library to combat drug ingredient counterfeiting. IPEC's David Schoneker told "The Gold Sheet" that he worries such a library could inadvertently create problems for legitimate drug ingredients.

Often, there is more than one way to make a pharmaceutical excipient, Schoneker said. A manufacturer might specify the use of an excipient made one way, while the library might contain a spectrum from a reference standard of the same excipient made another way. In such a case, the excipient in the manufacturer's product could appear to be counterfeit, even though it is not.

Emerging Solution in Search of Funding

Speaking in an anti-counterfeiting session of the conference, USP CEO Roger Williams said that with spectral libraries connected wirelessly to handheld analyzers, "you can see emerging a scientific solution to this."

Williams said he believes USP could establish a spectral library system in collaboration with the World Health Organization and others. However, he wondered who would offer to provide the $10 million or $15 million a year that he figured it might cost to maintain such a system.

If asked, he feared WHO, the World Bank, FDA and others might decline to fund it, directing him instead to the Bill & Melinda Gates Foundation. However, he said, "There must be a line stretching from Seattle into Montana for people who are trying to get money from Gates."

Comparable technologies are already in use in other fields, Williams noted. For example, he noted that the pharmacopoeia's chief metrology officer, Bill Koch, always travels with a Clear card, which by virtue of biometric fingerprint and iris imaging enables him to skirt long security lines at airports.

"If you can identify Bill Koch unequivocally," Williams asked, "why can't you identify ibuprofen unequivocally?"

- Bowman Cox ([email protected])