Agnes Shanley is senior editor of Pharmaceutical Technology.

Agnes Shanley

	Today, a growing number of compliance and quality failures can be traced to problems with API and excipient manufacturing. Finished drug manufacturers must be sure that the ingredients used in their products meet current good manufacturing standards, but the expense and time required for audits can be a significant burden. From the ingredient supplier’s perspective as well, having pharmaceutical customers conduct thorough audits of their facilities can pose serious challenges. One solution that the world’s regulatory agencies are supporting is having experts from independent authorities conduct these audits and provide certificates of compliance.

	Sessions at CPhI will discuss pharmaceutical and raw material quality in great depth from Oct. 9–11, 2018 in Madrid, Spain. For more information on third-party inspection and certification, this article from 

September supplement on pharmaceutical materials, written by experts affiliated with the International Pharmaceutical Excipients Council (IPEC) outlines how the practice works, what is required, and the benefits it offers (1).

	1. D.B. Klug, R.D. Lindblad, D.G. Muse, K. L. Ulman, P. Walsh, and P.S. Zawislak, 

Pharmaceutical Technology APIs, Excipients, & Manufacturing 2018 

Articles

Audits by independent experts promise to ease the burden that customer audits can have on suppliers, while ensuring drug manufacturers that ingredients meet strict cGMP standards.

How Third-Party Audits Can Help Ensure Excipient Quality

It is not unusual to hear of cGMP and quality failings in API and finished drug manufacturing, especially as more functions are outsourced. Between October 2016 and September 2017, out of 3343 citations for quality systems failures, roughly 11% were likely due to problems with supplier quality management, according to Phil Johnson, senior principal for quality and compliance services at IQVIA (1).

	But the root causes for some quality failures can be extremely difficult to sort out. This is becoming particularly evident in the valsartan recalls, which began in July 2018 after traces of the toxic nitrosamine, N-nitrosodimethylamine (NDMA), were found in the APIs used to manufacture generic sartans, the angiotensin inhibitor blockers (ARBs) prescribed to some patients to treat high blood pressure.

	By August 27, 2018, valsartan from 16 different suppliers had been yanked from pharmacy shelves (2). NDMA, classified as a “probable human carcinogen,” was found in API made by Zhejiang Huahai, a manufacturer in China. But subsequently, traces of another nitrosamine contaminant, nitrosodiethylamine (NDEA), were discovered in a batch of another ARB, losartan, made in India by Hetero Labs, and in lots of API made by Zhejiang Huahai and of generic valsartan distributed by Torrent Pharmaceuticals.

	NDMA’s toxic effects in animals have been known since the 1950s (3), and it was the poison of choice in two murders in 1978 (4). Over the past few decades, growing evidence of nitrosamines’ potential impact on human health has helped drive public area smoking bans and intensive process changes in the food industry (5,6). While investigations into the root cause of the contamination continue, this article touches on some questions that the case has brought up so far. Of particular concern is the way that the industry assesses process synthesis risks, especially for small-molecule APIs whose processes may generate trace levels of genotoxic impurities. Most of these compounds are manufactured overseas, but compendial testing requirements may not be enough to clue manufacturers into the need to monitor and test for trace levels of genotoxic contaminants. As offshoring and outsourcing trends continue, the recall suggests that developing solutions will be crucial.

	It is believed that NDMA contamination resulted from changes that Zhejiang Huahai made to its manufacturing process in 2011 and 2012, using a method that was patented in 2014 to reduce waste and improve product yield. Zhejiang Huahai had submitted documentation for the process change to regulators, and no objections were found. “The FDA and the European Directorate for the Quality of Medicines and Healthcare (EDQM) approved the changed process, but may have missed the potential for formation of genotoxic impurities,” says Philippe André, a cGMP auditor with Qualandre, based in Zhejiang, China, who inspected the Zhejiang Huahai facility.

	What began as a single case has snowballed into a major risk-assessment puzzle. The European Medicines Agency (EMA) is considering not only valsartan and losartan, but candesartan, irbesartan, and olmesartan in its efforts to find root cause (7). Both FDA and the General European Official Medicines Control Laboratories Network (GEON) published methods

	for testing for the impurities in August and September (8,9). FDA and EMA also found cGMP deficiencies at the company’s facility. FDA placed the company’s products under Import Alert on September 28 (10), when EMA also revoked its right to sell the product in Europe (11). In an FDA 483 published on September 21, based on inspections in July and August (12), FDA found fault with the company’s change-control system and its “failure to evaluate all potential risks from the 2011 manufacturing process change.” The company had hired an outside lab to conduct a small-scale research project assessment without pilot-scale testing or a formal risk assessment, the inspectors wrote. In addition, the 483 found that the company did not have a quality agreement in place with that outside lab. Inspectors also found fault with Zhejiang Huahai’s inconsistent classification of risks in different process change documents. Where the initial change request classified the process change as critical, DMF amendments sent in 2013 classified the changes as minor, inspectors wrote. Among other problems, FDA inspectors also singled out inadequate validation, cleaning procedures, analytical methods, sampling and testing, and equipment maintenance: However, observers see some of

	these observations as focusing more on procedural details rather than fundamental risk assessment problems. Many of the problems noted during FDA’s site inspection may not have led to the presence of nitrosamines in valsartan, says André.

	“If Zhejiang Huahai did not identify the need to develop a control strategy to reduce the new risks introduced with the optimized process, neither did regulators when they approved the process change,” he says, “and the manufacturer’s failure in this regard was just part of an industry-wide failure led by the regulators.”

	In response to the valsartan recall, André’s company is now conducting audits that zero in on the potential for any process to generate genotoxic impurities. So far, audits have found three problematic synthetic drug substances, says André. One of them is levocarnitine, synthesized from a probable carcinogen, epichlorohydrin. Depending on how it is synthesized, the compound may not only contain epichlorohydrin, but also traces of cyanide.

	The yield of synthesis is not great, André says, so it is difficult to predict whether a residue of unreacted epichlorohydrin might be carried over in the final product. Nevertheless, he asks, “Which impurities does the US Pharmacopoeia require testing for? Chlorides, sulfates, sodium and potassium, none of which is toxic at such levels.” He wonders how many manufacturers of this compound are even aware of the potential risk.

	Even the Chinese manufacturing process for acetaminophen (a.k.a. paracetamol) is a point of concern, says André, since one of the early intermediates is the probable carcinogen, 1-chloro-4-nitrobenzene. “We have audited most of the major Chinese manufacturing plants of acetaminophen, and found no evaluation of and no testing for 1-chloro-4-nitrobenzene at any of them,” he says. André sees a need for manufacturers and regulators to pay much closer attention to potential risks in the manufacturing process. “In the valsartan case, the focus was on control of the related substances of synthesis and other impurities above the reporting threshold (0.05% in the case of valsartan), rather than on the safety of the chemical synthesis processes.

	Zhejiang Huahai’s improved process replaced tributyltin azide with the more toxic compound, sodium azide, says André. As a result, the yield of tetrazole formation was much better. However, sodium nitrite was used to destroy the excess sodium azide that remained after the synthesis step. Sodium nitrite is often used as a decontaminating agent of sodium azide in acidic conditions, André says. However, under these conditions, it forms nitrous acid, which could react with the residue of dimethylamine in dimethylformamide, the solvent that is used in the tetrazole-forming reaction, to generate NDMA, says André. 

	“The possible formation of nitrosamines from nitrites and secondary amines in acidic conditions was already well-known to the food industry,” says André. “The use of sodium nitrite should have been a red flag prompting a check of possible presence of secondary amines, but it was not,” he says.

	“So we arrive at the million-dollar question: Are regulatory agencies and pharmacopeias doing a good enough job, if a sponsor can comply with [most] regulations and yet send a product on the market which contains carcinogens,” asks Anders Fuglsang, founder of Fuglsang Pharma. “We can’t test for everything, but I’m not entirely happy with that statement as a patient or consumer,” he says. Fuglsang hopes that there will be an independent analysis of the root cause of the nitrosamine contamination, performed by independent experts outside of regulatory agencies or pharmacopeias. In the end, he says, “we can only find what we are looking for.” But the sartan API contamination case suggests a need to focus more closely on assessing potential risks during process synthesis review.

	Preventing situations like this from occurring in the future will be complex, says Fuglsang, and require getting all the different players involved to see the bigger picture, from pharmacopeias and regulators, to finished drug manufacturers, API manufacturers, and national testing labs. “At this point,” he says, “that may be wishful thinking.”

	André sees the root cause study as an opportunity for the industry to look more deeply into the way it approaches risk assessment. This will be especially important for API syntheses that may result in residual levels of potentially genotoxic impurities. “I hope we will all draw the right lessons from this [recall] debacle,” says André. “Despite its mistakes and deficiencies, Zhejiang Huahai basically did what the regulators expected from them at the time. Stoning the company would be a distraction from the critical deficiency in the regulatory supervision of drug substances,” he says.

	1. P. Johnson, “Quality In, Quality Out:Regulatory Drivers of Supplier Quality Management in the Pharm and Biopharma Industries,” Xtalks Webcast (September 18, 2018).
	2. FDA, “

, August 28, 2018.
	3. P. Magee and J. Barnes, “British Journal of Cancer , 10 (1) 114-122 (1956),
	4. C. Purdy, “

A Common Blood Pressure Medicine is Being Recalled Because it Was Mixed With NDMA

 (1991).
	6. M. Jagerstad, Mutation Research 574(1-2), 156-172 (August 2005)
	7. EMA, “Valsartan: Review of Impurities Extended to Other Sartan Medicines,” Press Release, September 21, 2018.
	8. FDA, “FDA Announces Preliminary GC/MS Headspace Method for Determining NDMA in valsartan” Press Release, August, 2018.
	9. EDQM, “OMCLs Release Three Methods for Determination of NDMA in Sartans,” Press Release, September 21, 2018.
	10. FDA Press Release, “

FDA Places Zhejiang-Huahai on Import Alert

,” fda.gov , September 28, 2018.
	11. EMA, Press Release, “

EU Inspection Finds Zhejiang Huahai Site Non-Compliant

,” ema.europa.eu , September 28, 2018.
	12. FDA, 


	Vol. 42, No. 10
	October 2018
	Pages: 60–63

	When referring to this article, please cite it as A. Shanley, “Sartan Recalls Beg the Question: Is Compendial Impurity Testing Enough?” 

Experts blame the recalls, not on cGMP failures, but on inadequate risk assessment of processes that can generate toxic impurities.

Sartan Recalls Beg the Question: Is Compendial Impurity Testing Enough?

	Recalls of valsartan, the API found in Novartis’ blood-pressure treatment, Diovan, and its generic descendants, have underscored vulnerabilities in the way that the pharmaceutical industry assesses and manages global supply-chain risks. The recalls were initiated in the United Stated and European Union in 

, after N-nitrosodimethylamine (NDMA), was found in API used to manufacture generic angiotensin receptor blockers (ARBs). The nitrosamine, classified as a probable carcinogen, was found in active ingredient manufactured by Zhejiang Huahai Pharmaceutical Company, an API supplier in China, and a number of generic products using it were removed from pharmacy shelves (1). 

	FDA released analytical methods for detecting the contaminant in August 2018 (2). Later, regulators in Germany found low levels of another nitrosamine, N-nitrosodiethylamine (NDEA), in one batch of losartan manufactured in India by Hetero Labs. Zhejian Huahai had also detected NDEA in some lots of its valsartan API, and alerted FDA; subsequent testing by FDA confirmed its presence in several lots of valsartan API and in three lots of valsartan sold by Torrent Pharmaceuticals (3), that used that API. 

	As a precaution, to ensure that synthesis process changes or other factors have not resulted in systemic contamination of sartan drugs now on the market, the European Medicines Agency (EMA) will now expand its review of nitrosamine impurities to include four other widely-prescribed active ingredients: 

	candesartan, irbesartan, losartan, and olmesartan.

	As the agency explained in a press release announcing the effort on Sept. 21, 2018 all these compounds share a tetrazole ring structure (4). Different synthesis routes could lead to the formation of nitrosamine impurities during manufacturing.

	Also on Sept. 21, 2018, the General European Official Medicines Control Laboratories Network (GEON) published three methods for detecting the impurities. These analytical tests were developed in Ireland, France, and Germany by the Public Analyst’s Laboratory in Galway (PALG); the French OMCL in Montpellier and the German OMCL at the Chemisches und Veterinär-Untersuchungsamt (CVUA) (5):

		PALG’s method, based on headspace-gas chromatography and mass spectroscopy (GC-MS) (single quad), can be used to determine the presence of NDMA in APIs and sartan tablets

		The French method, based on high-performance liquid chromatography and ultraviolet (HPLC-UV) analysis, can be used to test both API and finished drug product

		CVUA’s method uses atmospheric pressure chemical ionization-ultra-high-performance liquid chromatography and mass spectroscopy (APCI-UHPLC-MS/MS) to both detect and quantify NDMA in drug products.

	FDA and EMA have emphasized that patients who continue to take medications that may contain these impurities face very low risks of developing serious health problems. In August 2018, FDA scientists said that it would take 8000 patients taking the maximum daily dose of the drug (320 mg) each day for four years, to result in one additional case of cancer (6). Regulatory agency officials have asked that patients remain on these medications until their physicians prescribe alternatives. 

	However, regulators must now grapple with the question of how members of a class of potent probable carcinogens wound up in the pharmaceutical supply chain in the first place.  NDMA, widely used in cancer research, was the poison of choice in two sensational murders in the U.S. and Germany (7) in 1978, while growing evidence of its impact and that of other nitrosamines helped drive public area smoking bans and intensive process changes in the food industry.  

	According to the World Health Organization, nitrosamines are not used industrially in the US or Canada, but may still be released as byproducts from municipal wastewater treatment facilities and some industrial manufacturing facilities (8). 

	Since research in the 1950s first revealed their carcinogenic effects on animals (9), regulators have worked to reduce potential exposure, and efforts have helped lead to tobacco regulation (10) and regional smoking bans at offices, restaurants, and other indoor public areas in many parts of the world. Food processors have studied their generation during heat- (11) and acid-catalyzed manufacturing, and come up with alternative methods to reduce or prevent their release.

	It is believed that the NDMA contamination could have resulted from changes that Zhejiang Huahai made to its manufacturing process in 2012, using a method that was patented in 2014 (12) to reduce waste and improve product yield.  

	Depending on which press report one reads, the presence of contaminant was either discovered by the manufacturer during process optimization, or by Novartis’ generic drugs subsidiary Sandoz, during routine raw material testing (13). 

	The company had submitted documentation of the process change to regulators, and no objections were found. “FDA and the European EDQM approved the changed process, but may have missed the potential formation of some genotoxic impurities,” says Philippe Andre, a cGMP auditor with Qualandre, based in Zhejiang, China, who inspected the Zhejiang Huahai facility.

	These issues would not have come up in a customer review of changes to the manufacturing process, either. So if the resulting problems reflect a failure of change management, it occurred at a fundamental level.

	Media reports of past warning letters and quality issues at Zhejiang Huahai’s facilities (14) may have obscured the fact that the manufacturer had followed all the legal requirements, and submitted all required documentation.  

	 “The inspections revealed some problems, none of them critical and all unrelated to the presence of NDMA in valsartan,” says Andre. 

	Zhejiang Huahai had no quality control or cGMP failures in this case. In fact, Andre says, the firm’s quality management was much better than average. “The contaminant was present and toxic at levels below the reporting limit for impurities specified in the pharmacopeia,” he says.

	Andre sees a need for manufacturers and regulators to pay more attention to potential risks in the manufacturing process. “In this case, the focus was on control of the related substances of synthesis and other impurities above the reporting threshold (0.05% in the case of valsartan), rather than on the safety of the chemical synthesis processes. Preventing such failure would require a more attentive assessment of the production processes,” he says.

	As Andre explains, in order to synthesize the tetrazole cycle in the valsartan molecule, Zhejiang Huahai’s improved process replaced tributyltin azide with the more toxic compound, sodium azide.  As a result, the yield of tetrazole formation was much better. However, sodium nitrite was used to destroy the excess sodium azide remaining after the synthesis step. Sodium nitrite is a well-known decontaminating agent of sodium azide in acidic conditions, Andre says. 

	Under acidic conditions, sodium nitrite forms nitrous acid, which could react with the residue of dimethylamine in dimethylformamide-the solvent that is used in the tetrazole-forming reaction-to generate NDMA, says Andre.

	Facility inspections cannot, and are not meant to, assess the synthesis process, says Andre; regulators must do that based on information provided by manufacturers. The risk of NDMA contamination could have been predicted during that assessment, he says.

	He suggests that the current situation may stem from focusing on the wrong priorities (i.e., controlling the related substances of synthesis) as required by the pharmacopeia, Andre says. 

	“The possible formation of nitrosamines from nitrites and secondary amines in acidic conditions was already well-known to the food industry,” says Andre. “The use of sodium nitrite should have been a red flag prompting a check of possible presence of secondary amines, but it was not,” he says.

	In this case, the API manufacturer and customers apparently followed the rules and standards set by regulatory agencies and pharmacopeias, says Anders Fuglsang, founder of Fuglsang Pharma. “So we arrive at the million-dollar question: Are regulatory agencies and pharmacopeias doing a good enough job, if a sponsor can comply with regulations and yet send a product on the market which contains carcinogens,” he asks. “We can’t test for everything, but I’m not entirely happy with that statement as a patient or consumer,” he says.

	Fuglsang hopes that there will be an independent analysis of the root cause of the nitrosamine contamination, performed by independent experts outside of regulatory agencies or pharmacopeias. In the end, he says, “we can only find what we are looking for.” But the sartan API contamination case suggests a need to focus more closely on assessing potential risks during process synthesis review. “All sorts of chemicals or residual impurities may never show up in the standardized testing mandated by pharmacopeias,” says Fuglsang.  

	Preventing situations like this from occurring in the future will be complex, Fuglsang notes, given the different players involved-from the pharmacopeias, which set standards for the APIs and excipients, to the regulatory agencies, whose guidelines refer to the pharmacopeias and standards for development, to finished drug manufacturers who buy API and excipients from subcontractors, to subcontractors who make the API, and the national testing labs, which test products to ensure that they meet standards.

	In the end, it will be crucial to get all these different stakeholders together to discuss issues and to see the bigger picture. At this point, Fuglsang says, “that may be wishful thinking.”

, “FDA announces voluntary recall of several medicines containing valsartan following detection of an impurity,” 

, “FDA Announces Preliminary GC/MS Headspace Method…” August 2018.
	3. J. Christensen, "

FDA Finds Additional Impurity in Heart Drug

, September 15, 2018, https://www.cnn.com/2018/09/13/health/valsartan-recall-additional-impurity-fda/index.html
	4. EMA 

, “Valsartan: Review of Impurities Extended to Other Sartan Medicines,” 

, “OMCLs Release Three Methods for Determination of NDMA in Sartans,”

, September 21, 2018.
	6.  J. Christensen, “

FDA Expands Recall of Blood Pressure Drug

Concise International Chemical Assessment Document 38: NDMA

,” World Health Organization,Geneva, 2002.
	9. P. Magee and J. Barnes, “

Production of Malignant Primary Hepatic Tumors in the Rat by Feeding NDMA

, 10(1) pp114-122 (1956).
	10. A. Tricker 

NItroso Comounds in Cigarette Tobacco and Their Occurrence in Mainstream Tobacco Smoke

12(2); 257-261 (1991).
	11.M. Jagerstad, “Genotoxicity of Heat Processed Foods, Mutation Research 574(1-2), 156-172 (August 2005).
	12. Patents Issued to Zhejian Huahai Pharma Co., Ltd, 

www.patents.justia.com/assignee/zhejiang-huahai-pharmaceutical-co-ltd
	13. A. Harney, B. Hirschler, “

Toxin at the Heart of Drug Recall Shows Holes in Medical Safety Net

US FDA Flagged GMP Concerns at Valsartin Manufacturing Plants in Two 483s

The European Medicines Agency’s detection of a second nitrosamine in a sartan API is driving a deeper dive into tetrazole chemistry; root-cause investigations will now include not only valsartan and losartan, but candesartan, irbesartan, and olmesartan.

After Valsartan Recalls, Regulators Grapple with Nitrosamine Contamination in APIs

	In July 2018, The Tufts Center for Drug Development released a report (1), which found that more complex clinical trial protocols are driving up clinical trial costs. Examining 9737 protocols from 178 global pharmaceutical and biotechnology companies, Tufts researchers found that Phase III trials have seen the greatest increase in complexity over the past 10 years, while companies are collecting data on 86% more endpoints than they did last decade, necessitating new procedures. 

	In addition, Tufts found, between the four-year periods 2001–2005 and 2011–2015, pharmaceutical manufacturers doubled the number of countries involved in clinical trials, and increased the number of Phase III support sites by 63%.

	For operating companies and their contract partners, whether contract development and manufacturing or contract research organizations (CDMOs or CROs), the increased complexity has posed some serious challenges. Impacts are being felt on the manufacturing and operations side, which, according to recent estimates, already accounts for 20% of all clinical trial costs, or some $30 billion per year (2). 

	Manufacturing is feeling the impact of increased protocol complexity in terms of risk mitigation (i.e., in the way that studies are being designed), as phases are combined or changed in midstream based on interim results.This approach requires far more focus on upfront planning than traditional study design did. “Maintaining the flexibility and agility needed to react to changing clinical protocols is a very real challenge,” says Mark Lewis, Alcami site director of Wilmington Drug Product Manufacturing Operations.

	“Client requirements in early phase studies often evolve based on clinical results. This frequently means redevelopment or expansion of the dosage range covered by manufacturing,” says Lewis, “And our ability to respond to changes that benefit the patient is critical.” Any shortages of investigational medicinal products (IMPs), or other delays, can affect results of a trial and lead to financial losses of $1 million per day (2).

	Driving much of this change is the move to personalized medicine, says Andrea Zobel, senior director of product management, clinical trial supplies and logistics, for PAREXEL. Replacing the “one-drug-fits-all” model are individual preparations and dosages, and patient-specific drugs in autologous therapies. Consequently, patient numbers per batch are decreasing, and manufacturing must be flexible for products with short shelf-lives. 

	In addition, Zobel says, clinical materials are being manufactured closer to the clinical site, and the timelines for GMP quality processes are being shortened and automated. “Manufacturing sites with partial automation to produce hundreds or thousands of packages are no longer required. Instead, manufacturing sites with many parallel workstations are needed to produce many small batches continuously,” she says. 

	She cites an example of a pharmaceutical firm whose focus has shifted from cardiovascular therapies to oncology drugs. “In oncology today, the majority of innovative drugs are biopharmaceuticals with short shelf-lives and requirements for cooled or frozen storage and distribution. Therefore, the decoupling point when a drug package is made to be trial-specific should be as late as possible.” 

	Currently, Zobel explains, the manufacturing site for clinical supply of all global studies is comparably small, with approximately 20 different labeling/packaging rooms. These rooms are continuously producing clinical trial medication for many parallel trials with various treatment arms.

	The most advanced process is “manufacturing on-demand,” an approach in which the IMP kit is labeled just-in-time for when the shipment request for a site is generated by an IRT system.While the process itself can be supported by technical solutions such as label printing on demand, label work stations connected to IRT systems, and decentralized manufacturing sites with standardized equipment, GMP-compliant quality controls and releases remain a challenge. Again, technical solutions can help to give the quality managers the required oversight and documentation to enable even remote releases (e.g., in-process-controls using bar-coded components), which are tracked in an electronic release system and can replace the paper based and visual checks. 

	Additionally, the regulations must be adapted for some new therapies, especially cell and gene therapies, which are prepared individually for single or small groups of patients and cannot comply with the traditional quality and analytical process steps. Regulations such as the EU guidelines on GMP for Advanced Therapy Medicinal Products (ATMPs) address these new requirements, says Zobel.

	Interactive response technology (IRT), including interactive voice response (IVR) software, first appeared on the scene about 20 years ago to help sponsors and contract partners better respond to clinical trial complexity. The software aims to help users manage increased randomization and complexity, and help ensure that drugs are distributed to trial sites and dispensed to patients correctly.  

	Cenduit, for instance, was launched in 2007 as a joint venture between Fisher Clinical Services (now Thermo Fisher Scientific) and Quintiles (now IQVIA) to help address these problems.  Since that time, eClinical IT systems have been developed, and IRT programs are interfacing with electronic data capture (EDC) systems and distribution systems, and with lab systems or electronic clinical outcome assessment (eCOA) tools to improve clinical trial manufacturing and operations.

	Still, it is up to sponsors to ensure that data are adequate and properly interpreted to optimize inventory levels and delivery times. “Complex study protocols with dose changes, re-randomization, and dynamic design updates are much harder to plan for with regards to how much drug is needed. Initial planning needs to be done based on numerous assumptions, and it’s important to continuously fine tune the forecast of drug needs during the study based on actual data. This approach increases planning certainty and accuracy and allows a reduction in expensive overages,” according to Stefan Dürr, senior director of client delivery and head of Cenduit’s Drug Supply Center of Excellence. 

	Anticipating the impacts of clinical trial protocol changes means that manufacturers face moving targets. Often, manufacturing and packaging must start even before study portfolios are finalized, says Dürr. 

	As a result, sponsors often gear manufacturing to worst-case scenarios to ensure adequate supplies, but the result is wastage, and, with costly biopharmaceuticals, in particular, the economic impact can be high. In reviews of more than 200 clinical trials, Cenduit has found that 65% of the clinical supplies packaged for those trials were not used at the end of the study. 

	Some manufacturers are using modeling tools to help improve initial project planning. Used with IRT, simulating for forecasting and management offers insights into study designs and supply strategies that allow avoidance of excess inventory and optimize overage, considering possible delays in shipping (e.g., customs), says Vince Santa Maria, vice-president of worldwide clinical trial material and depot operations at PAREXEL.

	Simulation, with IRT, can also be done during a study, using actual study IRT data to help mitigate issues along the way as well as up front, says Santa Maria.  “In such cases, overages are discussed with the sponsor early on during the setup phase of a study to establish an optimal approach later on. Risk mitigation strategies and suitable potential remediation can also be discussed at length during simulation to assure for proper planning upfront before issues occur, as well as over the lifetime of the study as patterns emerge,” he says. 

	Software such as Bioclinica’s Monte Carlo simulation software, Optimizer, can be used to develop different forecasting and supply scenarios, based on how the studies might be run. In addition, contract services providers are becoming creative in setting up and maintaining the IRT’s resupply module, says Melissa Peirsel, Manager of IRT Services at Sharp Clinical Services. “Resupply has never been a ‘set-it and forget it’ portion of a clinical study, where we could once set the resupply parameters for the entire study based on where the study was with recruitment, we are now finding that the system needs to have specific resupply settings for the demand at each site.” When it comes to global trials, she says, there is a growing requirement for ancillaries to be sourced locally. 

	“We have also been seeing a trend where there are more kit types, but less supply available to support the trial. This places a demand on the IRT system to have just the right amount of trial materials on site without causing insufficient inventory errors or overstocking a site that will never use the supply. Some sponsors are resorting to site-to-site transfers of materials and evaluating returned unused materials for redistribution to other study sites, Peirsel says.

	Planning relies on sharing data more effectively, and more sponsors are partnering with contract partners to ensure that clinical supplies are delivered on time and in full. In addition to simulation tools, Santa Maria sees a move to greater use of just-in-time (JIT) labeling, in which medication is initially labeled without protocol or country- specific information, without a batch number or expiry date, and with the minimal information to ensure correct picking against a shipment request. Complete regulatory-compliant labeling is then performed as close in time to shipment to the site as possible. Final packaging, labeling, release, and shipment are performed within 48 hours of receipt of trial/country/patient specific shipment request.

	IRT can also help prevent quality issues, says Dürr, by creating the kit list required for labeling and excluding expiry dates from the label if it is controlled by IRT. “This approach can reduce drug wastage and remove the requirement to relabel drugs at depots and sites if stability allows to extend the expiry date,” he says.

	Through collection of additional data, the risk of patients taking a therapy after its expiration date could be reduced even further, Dürr notes. For example, sites could scan the kit number prior to dispensing and when the patient returns the drug, so that IRT can check that the drug was indeed returned prior to the expiry date, and send alerts in case there is a drug with the patient that is going to expire soon.

	Within the past five years, returns management, or the need to return unused experimental treatments, has become more of an issue. Compliance requires closer connections between partners and sponsors, says Lewis, and companies are recording data and functionality into their IRT systems. 

	Even though returns and destruction of unused IMPs are required, and part of all good clinical practice (GCP) and GMP regulations, follow through is often an issue in clinical trials, says Zobel. “Processes are often not described in enough detail during study set-up; training is often not sufficient ;and responsibilities are often not assigned,” she says. 

	Zobel believes that the upcoming EU Regulation (EU) No 536/2 (3) will add additional complexity when the tracking and documentation is expanded to used and unused IMPs and auxiliary products. However, she notes, new requirements will also improve clarity because they will require description in the protocol of the specific arrangements for tracing, storing, destroying, and returning IMPs.

	Although the clinical site is initially planned to be where the IMP is destroyed, many trials’ country regulations and site capabilities often require a reverse supply chain with return shipments to certified destruction facilities in the same or other countries. Sponsors must consider carefully whether a network of local depots or a central hub will be the best and cost- effective solution, Zobel says. 

	Recently, IRT systems are being used to handle reverse logistics (e.g., with reports of used or unused IMPs and automated return shipment requests), says Zobel. These systems also support clinical site accountability. “The new trend to apply barcodes on IMP packages, instead of reading and entering kit numbers, allows packages to be scanned and tracked automatically and reduces workloads for clinical staff while improving data quality,” she says.

	Ensuring cold chain performance has become essential to ensuring patient safety, says Santa Maria. Close integration between IRT and distribution systems is essential to enable manufacturers to manage supply risks faster and ensure there are no site stock outs, says Dürr. Thorough testing is required to ensure that IMPs are kept at the right temperature, not just in transit but also during storage.  

	The move away from traditional trial models to direct-to-patient (DTP) approaches is also complicating clinical trial operations. This newer approach to clinical trials may make logistics more complex and costly, says Dürr.  However, he notes, it will also reduce inventory requirements at clinical trial sites, make trials more patient-centered, and reduce overall IMP oversupply problems. 

	The new DTP models will also move manufacturing to the pharmacy level, says Zobel. At this point, CMOs are only licensed in a few countries to handle this type of manufacturing and have a licensed pharmacist available to dispense medication directly to patients. In addition, most pharmacies have no GMP licenses for IMP primary and secondary packaging, she says. 

	As the move to personalized therapies and DTP models continues, clinical trial complexity will remain a “given” for drug manufacturers.  IRT will continue to play an important role in ensuring continuity of clinical trial supply to sites, while also minimizing drug overage and wastage, predicts Tim Gilbert, senior director of product management, randomization, and trial supply management at PAREXEL.  

	1. Tufts Center for the Study of Drug Development, 

,“Rising Protocol Complexity Is Hindering Performance while Driving Up Cost of Clinical Trials,” 

Why is There So Much Wastage in Clinical Supplies?

” May 31, 2018.
	3. European Commission, EU No 536/2014, 

Clinical Trials Regulation EU No. 536/2014


	Vol. 42, No. 9
	September 2018
	Pages: 60–63

	When referring to this article, please cite it as A. Shanley, “Clinical Manufacturing: Clearing Higher Hurdles” 

Increasingly complex trial protocols have added to IMP manufacturing challenges.

Clinical Manufacturing: Clearing Higher Hurdles

Few global supply chains are as complex and fragmented as the pharmaceutical supply chain. Medicines flow, not only between manufacturers and a huge number of distributors, but between pharmacies and hospitals, and from manufacturers to pharmacies (1). Adding to that complexity is the challenge of ensuring that the quality of active ingredients and raw materials used to make finished drug products, most of which are made outside the United States and Europe.

	As a result, the world’s consumers are increasingly vulnerable to fake, substandard, and illegally diverted medicines. A number of reminders have underscored that vulnerability.

	In November 2017, the World Health Organization (WHO) issued new statistics on the impact of drug counterfeiting and substandard manufacturing, based on estimates from its Global Surveillance and Monitoring effort, which began in 2013 (2). According to WHO, one in 10 pharmaceuticals currently sold in developing nations is either fake or substandard. Based on modeling studies, WHO reported, each year, 72,000 to 169,000 children may die from pneumonia and another 64,000 to 158,000 in sub-Saharan Africa alone may die from malaria. Extrapolating these numbers beyond these two types of therapies, and that one region, suggests the enormity of the problem. 

	In July 2018, in the US, lots of the hypertension and heart failure treatment, valsartan (3), were recalled after the carcinogen N-nitrosodimethylamine was found in product manufactured by an API supplier in China. As a result, six manufacturers that used the API voluntarily recalled some medications containing the ingredient, and the recall’s scope was gradually extended to include additional medications. In August 2018, FDA issued suggested methods for using gas chromatography to detect the contaminant (4). 

	In July 2018, the Pharmaceutical Security Institute (PSI) released results of its study (5) on pharmaceutical supply chain crime. “Globally, in 2017, we saw the highest number of pharmaceutical crime incidents ever documented,” says PSI director Tom Kubic. As in previous years, Kubic noted, illegal diversion, including the movement of unapproved medicines from overseas into more developed markets, was up by 14%, while counterfeiting was up by 7%. In addition, he notes, more than 1500 different types of medications were counterfeited, diverted, or stolen, up by 21%. “Traffickers are targeting more expensive therapies,” Kubic says.

	Pharmaceutical manufacturers are moving forward with serialization and aggregation efforts. They are also sharing more data (e.g., laboratory test data on samples bought from illegal online pharmacies) with law enforcement agencies, which Kubic sees as an important and positive development.  

	Even as these efforts progress, however, those at the very end of the pharma supply chain remain at risk. David Aguero, manager of medication systems and informatics at St. Jude’s Children’s Hospital in Memphis, TN, provided some insights at a June 2018 conference on serialization innovation sponsored by the technology provider, Adents (6). The hospital had begun stocking a specific transdermal medication patch, the latest version of which looked quite different from what its pharmacists were used to seeing. A pharmacist alerted him, and concerted efforts began to trace the source of product and specific changes. 

	Despite careful recordkeeping and documentation, considering the fact that 14,000 pharmaceutical products are in use at the hospital’s pharmacies, authentication proved to be a major challenge. In the end, Aguero said, it took one month to figure out where the patch was coming from.

	Among pharma’s top supply chain problems, observers say, are lax penalties for those caught counterfeiting or selling diverted materials. “Only a few countries in the world have ratified the Medicrime convention, which calls for jail time for counterfeiting,” notes Richard Bergström, external Pharma lead for SICPA SA, a privately-owned Swiss company whose inks are used for global currency and anticounterfeiting. The company has since diversified into other areas including overt and covert anticounterfeiting technology, including packaging and taggants. 

	“Criminality is product agnostic, whether you are trafficking in purses, people, or medicines,” Bergström says. “And at this point, one can still send two packages of drugs worth over $10,000 each into the supply chain without tracking them.”

	PSI has seen a trend continue in which higher priced medications are increasingly being counterfeited, diverted, and trafficked into developed nations, Kubic says, where, in developing nations, generic antimalarials and antibiotics remain the focus. 

	Other fundamental challenges are the vastness, profitability, and fragmentation within pharma distribution, particularly in developing nations, says Bergström. He cites statistics from a 2016 report by the World Bank’s financing arm, the International Finance Corporation (IFC) (7). “If the global pharma market is worth a trillion dollars, the total markup in world distribution is easily worth another trillion, and two thirds of that business occurs in emerging markets,” he says.

	“The US is extremely efficient at distribution, and Europe is also efficient. So where is the rest of the money going?” he asks. The IFC expects solutions to the pharmaceutical diversion problem to come from private companies, Bergström says.

	SICPA has been working with global governments on serialization programs for alcohol and tobacco products, and Bergström notes that these efforts are well established. African nations have control over these supply chains, he says, while they continue to struggle with pharmaceutical couterfeiting.

	Technology promises to create transparency in the supply chain, and to help pharma overcome some criticism that it does not permit global access to drugs. Despite public outcry over drug prices, some manufacturers have reduced costs significantly in developing nations to increase access, he says. These initiatives have often gotten bogged down in the distribution chain, and failed. “Remove the middlemen, and improved global drug access can happen,” he notes.

	Law enforcement has scored some recent victories in improving pharmaceutical supply chain security. In 2017, PSI saw a step-up in enforcement efforts, with a 10% increase in pharmaceutical supply-chain criminal arrests throughout the world.

	Although the International Criminal Police Organization’s (INTERPOL’s) major anticounterfeiting effort, Pangea, will not start until the autumn of 2018, Kubic notes, enforcement efforts have been making a dent, especially in Europe, China, and Latin America. However, Kubic notes, discussions are underway about how to improve the impact of enforcement efforts. For instance, in 2018, concerted enforcement efforts shut down the illegal website, canadadrugs.com (8). Kubic wonders how lasting the results will be. In some previous cases, he notes, including one involving online sales of fraudulent Norvasc high-blood-pressure treatments, the individual behind the websites that were shut down is once again actively involved in the illegal online pharmaceutical distribution scene.

	In the US, Kubic says, FDA and the US Drug Enforcement Administration (DEA) have become more engaged in addressing illegal fentanyl manufacturing and supply, an area where illicit drug manufacturing and counterfeiting intersect. This has increasingly become a “homegrown manufacturing” problem, with imports of the finished product decreasing, and more individuals buying tablet presses and manufacturing equipment.

	Efforts in serialization continue, although results have been mixed. TraceLink’s 2018 study found that only one-third of stakeholders were ready to meet requirements and that only 8% of manufacturers had integrated programs with their contract partners (9). In addition, European manufacturers remain behind the curve (10) in submitting data to central data hubs.

	Bergström was director general of the International Federation of Pharmaceutical Manufacturers and Associations (EFPIA), based in Europe, from 2011 through 2017. He joined SICPA in 2017, to help direct its serialization efforts. Europe’s approach to serialization has, from the start, been different from that in the US, whose roots were in efforts by the state of Florida to develop drug pedigrees, which led to pioneering work by Big Pharma companies. “Europe’s serialization has been driven by pharma companies, wholesalers, and pharmacies, which wanted a system for all of Europe,” he says.

	He sees capital costs as an obstacle to progress with serialization. “In Europe, there are many small- to mid-sized companies, particularly in the south and east, for which even a $200,000-line investment would be difficult,” he says. Bergström also notes some fundamental IT issues. At this point, he says, most production lines are being equipped with equipment for printing and verifying barcodes. “The challenge comes with sending all the data,” he says. “Aggregation is approached differently in the US and Europe. In Europe, on paper at least, companies are expected to send a long string of numbers to a central hub. In the US, you are supposed to aggregate, to send serial numbers and aggregated format to supply-chain partners,” he explains.

	The problem that pharma companies can run into in Europe, Bergström says, is that, even if they are large and have consolidated financial reporting systems, their supply-chain systems are usually maintained separately. “Often, each is also in a different type of IT environment,” he notes, “where the regulations call for reporting from one single gateway.” 

	Equipment-wise, many companies are ready, but data-transmission issues have made it difficult to comply; more than 22,000 companies are required to send their data but only a fraction of that number have done so so far (10).

	In addition, Bergström notes, although cloud-based systems are gaining ground, many local governments want serialization data to be based locally. “There is also an unwritten requirement that services and equipment be sourced locally,” he says. “It’s not so much a technical issue, as it is a question of political perception.”

	One question for pharmaceutical manufacturers, regulators, and for governments, is how serialization data and IT might be used in the future, whether for operational excellence or other initiatives. Bergström notes that European regulators developed serialization regulations with a view to using the IT infrastructure, eventually, to help monitor drug safety and effectiveness. According to this model, patients would eventually help collect the data.

	“That’s the future. If you can authenticate on top of serialization, aggregation, and traceability, you can be sure that each patient received the right drug at the right time, and then generate effectiveness data on top of that,” he says. This will become even more important as the industry moves to direct-to-patient clinical trials and into personalized therapies, he says.

	In the short term, interoperable regulatory databases help improve pharma security, he says, and serialization, container tracking, and other solutions could be applied to help ensure the pedigree of APIs.

	Blockchain could prove most useful, Bergström says, in establishing provenance for data, in clinical trial data management for medications and samples, and, downstream, in payment modules and monitoring patient outcomes in connection with payments, a requirement of value-based contracts.

	A growing number of authentication solutions are focusing on the end of the supply chain. Taking this approach to its endpoint is Valisure, which introduced a Raman-based analytical system coupled with algorithms to address variability in medication quality. This system could detect substandard API or insufficient API loading, in branded medicines, as well as generics, which currently fill 90% of the prescriptions in the US. CEO David Light established the company after a colleague suffered side effects when he refilled a prescription and was issued a drug from the same manufacturer. The technology has received ISO 17025 accreditation, and the company, which has partnerships with leading distributors, plans to launch an online pharmacy that would use the technology to ensure that drugs meet stated content requirements. The venture would initially focus on anticonvulsants, hypertension treatments, and antidepressants.

2016 Global Drug Safety and Traceability Report

, May 26, 2017.
	2. World Health Organization 

, “One in Ten Medical Products in Developing Countries is Fake or Substandard,” November 28, 2017, 


	3. FDA, “FDA announces voluntary recall of several medicines containing valsartan following detection of an impurity,” 

FDA Announces Preliminary GC/MS Headspace Method…


	6. D. Aguero, “Serialization: A Pharmacists’ Perspective,” a presentation at the Adents Serialization Innovation Summit, June 20, 2018.
	7. “Private Sector Pharmaceutical Distribution and Retailing in Emerging Markets,” International Finance Corp. and Ministry of Foreign Affairs, the Netherlands, 2016.
	8. “

Popular Online Pharmacy Shuts Down After Selling Counterfeit Cancer Drugs

, June 1, 2018.
	11. A. Shanley, “Serialization: Scaling Down for the Final Stretch,” 


	Vol. 42, No. 9
	September 2018
	Pages: 18–21

	When referring to this article, please cite it as A. Shanley, “Focusing on the Last Link” 

As counterfeiting, API manufacturing issues, and illegal diversion increase vulnerability, could dispensers and even patients play a greater role in securing the pharma supply chain?

Focusing on the Last Link

Content uniformity, which ensures that API is present in the right quantity in solid dosage forms, both within and between batches, is crucial to maintaining patient safety. Variability in API loading can become a safety problem when patients split tablets, a trend that has become common a way to reduce the cost of prescription drugs (1).

	Content uniformity testing requirements, however, may not always be clear, and the process may be taken for granted, which can result in quality problems. Technology is available to automate more of the sample preparation portions of the test, which tend to be the most error-prone. Ishai Nir, small-molecule products manager at Distek, Inc., describes challenges with 

, and discusses a new automated sample preparation system that was recently introduced to the market.

	Content uniformity testing often overlooked

: You have often described content uniformity as ‘the overlooked test.’ Given the predominance of generic drugs today, why would such an important quality-control test be overlooked?

: Content uniformity, which, technically, is part of the 

) Chapter <905> Uniformity of Dosage Unit (2), is definitely the red-headed stepchild of solid-dosage form testing. For one thing, it is one of the tests where the least detail or guidance is provided on execution. 

	For example, the entire section of Chapter <905> that covers the procedure for performing the content uniformity (CU) measurement reads, simply: ‘Assay 10 units individually using an appropriate analytical method.’ That’s it! Compare that with Chapter <711>, the dissolution chapter, which uses more than twice the number of words just to describe the specifications of a paddle, one of seven pieces of equipment that are required for dissolution testing. In 2018 alone, Goggle Scholar currently includes over 5500 references for dissolution testing, yet less than 1300 for content uniformity testing (2).

:  What impact does this lack of definition most often have on results?

: We haven’t seen the same proliferation of instrumentation dedicated to CU testing that we’ve seen in virtually every other required pharmaceutical test, and this is due to lack of guidance. Instead, companies have developed and validated their own methods based on existing general-purpose laboratory equipment such as stirrers, sonicators, homogenizers, and shakers.

	You can think of CU testing as having two components. The first is sample preparation. This entails finding a way of reproducibly extracting 100% of the API out of the dosage form into a solution, and, if necessary, adjusting the concentration of this solution via dilution so that it is suitable for the analytical finish. The solution is then analyzed, typically using ultraviolet (UV) or high-performance liquid chromatography (HPLC), to determine the API concentration. The concentration is then multiplied by the total dilution volume to calculate the original amount of API present in the dosage form.

	Most labs have multiple technicians on hand who are highly trained in measuring the concentrations of a given solution using UV analysis or liquid chromatography (LC). As a result, the analytical step has never been the challenging part of the CU test. Problems have generally arisen in sample preparation.

: What types of problems typically show up during CU sample preparation?

: There are two major sources of error in CU sample prep. The first is incomplete API extraction, and the second is dilution errors. Incomplete extraction takes several forms. One major source of variation is due to the fact that, as previously mentioned, unlike dissolution or disintegration testing, these procedures are not performed using a specific apparatus. This task is normally accomplished using non-analytical lab equipment such as magnetic stirrers or sonicators. Because the speed, frequency, or power of these units are not rigorously controlled, these parameters can vary over the lifetime of a unit, and between different units in one lab.  

	Variation can be even greater between units in different labs. So, if a method calls for sonicating the sample for six hours, that may be equivalent to four hours on one sonicator and eight hours on another. Often, method developers ‘solve’ this problem by calling for an overnight run to avoid these issues.

	The second source of errors is the dilution steps. Frequently, the 100% extracted solution is too concentrated to be read directly by analytical instruments and requires one or multiple dilutions. Of course, any time a processing step is added, a greater probability of operator error is introduced.

	In the case of sonicating samples, mistakes can be even more likely, because prolonged sonicating will normally heat a solution. If the measurements are taken and dilution performed before the solution has returned to room temperature, it can be easy for a lab technician to make a volumetric measurement mistake due to thermal expansion of the liquid.

: Why can’t the entire test be automated? Aren’t solutions already available to do this?

: A few systems have attempted to automate the entire CU process, including both sample prep and analysis. However, they are very expensive, starting at over $100,000 and ranging up to several hundred thousand dollars. In addition, they can be complicated to operate, difficult to validate, and expensive to maintain and keep running. In addition, although they address sample preparation concerns, they also attempt to address analysis, which never was much of a problem to begin with.

	The PrepEngine, developed by Distek in conjunction with Two Square Sciences, automates only the sample prep portion of the CU test. The goals behind its development were to make the extraction process considerably faster and more reproducible, and, to the extent possible, to eliminate the need for dilution steps entirely. It was also designed to be considerably less expensive than previous solutions.

: The CU sample prep system uses custom plastic vessels, designed for multiple uses, with a built-in blade and filled with appropriate media to extract the API from the dosage form. The main unit comprises 10 positions (to match the 10 samples required for a CU test), the control electronics, and user interface that allows programming the method parameters.

	The several-use (typically five runs) tubes are made of polypropylene, and are designed with internal finning, which helps redirect the dosage form down towards the blade as it is being broken apart and blended into the media. They are available in different volumes to match the concentration required for subsequent analysis without further dilution, whenever possible. 

	As a result, the operator using the equipment must only make one precise volumetric measurement of room temperature media during the entire measurement process. Finally, the tubes are available in amber for light-sensitive products to avoid sensitization before analysis.

	Variability in API loading can be a problem, especially with generic formulations. Improving content uniformity testing offers a solution.

: The situation is similar to that of disposable razors. Regardless of what material one uses to make the blades, they will become duller over time and require more agitation time to guarantee complete extraction. 

	Extending the sample prep time excessively to ensure complete extraction would not be the best solution. Instead, it is better to limit the number of samples that are ground using a given blade. 

	Some users have taken this concept to the limit, using each tube only once. In this case, the cost of the extra consumables is offset by the elimination of the time and cost associated with cleaning the tubes. This also eliminates the possibility of any product carryover due to inadequate cleaning. Additionally, using each component only once enables pre-sterilization of the tubes for biologics use.

: How complicated is method development using the CU Sample Prep Station?

: Method development using the CU Sample Prep Station is simple because analysis and calculation of the results remains largely the same as existing methods. The only exception is generally eliminating the need to correct for dilutions, which become unnecessary. 

	The sample prep portion of method development involves optimizing two parameters-agitation speed and duration. This requires conducting CU testing at different speeds and durations and extraction efficiency. The goal is to find the lowest agitation speed that will still guarantee complete release in an acceptable test time.

Sample preparation-specifically, API extraction and dilution-can introduce errors. Use of best practices and automation can reduce variability.