Abstract
Microbial quality control of non-sterile drug products has been a concern to regulatory agencies and the pharmaceutical industry since the 1960s. Despite being an old challenge to companies, microbial contamination still affects a high number of manufacturers of non-sterile products. Consequences go well beyond the obvious direct costs related to batch rejections or product recalls, as human lives and a company's reputation are significantly impacted if such events occur. To better manage risk and establish effective mitigation strategies, it is necessary to understand the microbial hazards involved in non-sterile drug products manufacturing, be able to evaluate their potential impact on final product quality, and apply mitigation actions. Herein we discuss the most likely root causes involved in microbial contaminations referenced in warning letters issued by US health authorities and non-compliance reports issued by European health authorities over a period of several years. The quality risk management tools proposed were applied to the data gathered from those databases, and a generic risk ranking was provided based on a panel of non-sterile drug product manufacturers that was assembled and given the opportunity to perform the risk assessments. That panel identified gaps and defined potential mitigation actions, based on their own experience of potential risks expected for their processes. Major findings clearly indicate that the manufacturers affected by the warning letters should focus their attention on process improvements and microbial control strategies, especially those related to microbial analysis and raw material quality control. Additionally, the WLs considered frequently referred to failures in quality-related issues, which indicates that the quality commitment should be reinforced at most companies to avoid microbiological contaminations.
LAY ABSTRACT: Microbial contamination of drug products affects the quality of non-sterile drug products produced by numerous manufacturers, representing a major risk to patients. It is necessary to understand the microbial hazards involved in the manufacturing process and evaluate their impact on final product quality so that effective prevention strategies can be implemented. A risk-based classification of most likely root causes for microbial contamination found in the warning letters issued by the US Food and Drug Administration and the European Medicines Agency is proposed. To validate the likely root causes extracted from the warning letters, a subject matter expert panel made of several manufacturers was formed and consulted. A quality risk management approach to assess microbiological contamination of non-sterile drug products is proposed for the identification of microbial hazards involved in the manufacturing process. To enable ranking of microbial contamination risks, quality risk management metrics related to criticality and overall risk were applied. The results showed that manufacturers of non-sterile drug products should improve their microbial control strategy, with special attention to quality controls of raw materials, primary containers, and closures. Besides that, they should invest in a more robust quality system and culture. As a start, manufacturers may consider investigating their specific microbiological risks, adressing their sites' own microbial ecology, type of manufacturing processes, and dosage form characteristics, as these may lead to increased contamination risks. Authorities should allow and enforce innovative, more comprehensive, and more effective approaches to in-process contamination monitoring and controls.
Introduction
Microbiological contamination of non-sterile drug products is an old challenge to the pharmaceutical industry. The consequences of this problem are quite serious and go beyond the direct costs involved in batch rejections or product recalls, as microbial contaminations represent a major risk for patient's health and may even be life threatning, not to mention damaging to a company's own reputation (1).
Currently, microbiological contamination is one of the top ten causes for drug product recalls in the US and UK markets. According to Grady (2) at least once per month, clinically relevant drug products are recalled from the US market, most of them distributed nationwide or beyond US borders.
Sutton and Jimenez (3) reviewed all recalls related to drug products and health care products registered by the US Food and Drug Administration (FDA) database between the years 2004 and 2011. During that period 642 recalls were recorded involving microbiological incidents, of which 22% concerned non-sterile drug products. The high number of recalls related to microbial contaminations of non-sterile drug products may be an indication of insuficiently defined or poorly implemented microbiological control programs (4, 5).
To better understand the microbiological pattern and define what can be done, it is necessary to know the general microbial hazards involved in a drug product's life cycle. This sort of information is not easy to obtain, and most of the companies do not know where and how to start. A valuable source of information are the warning letters (WLs) and non-compliance reports issued by health authorities (HA). Those documents list violations of regulatory significance identified by auditors during their inspections or investigations. A closer examination of these documents may reveal failures directly or indirectly related to microbial contamination (6, 7).
With such data and domain knowledge at hand, the next step is to use a formal QRM approach to better exploit those WLs (8). A QRM plan will allow risk identification, quantitation, and prioritization, followed by implementation of mitigation actions. QRM is an effective way to ensure quality and safety of medicines over their manufacturing life cycle (5).
In this context, the present work was designed to assess the main occurrences and good manufacturing practices (GMPs) failures related to potential root causes for microbiological contaminations in non-sterile drug products reported in WLs. The second step is to priotize the criticality of possible risks and of corresponding mitigation actions by applying a formal QRM approach. QRM principles provide manufacturers with a foundation to prioritize mitigation actions and to justify decisions leading to better controls and overall robustness in regard to contaminations.
Quality Risk Management
The GMPs for the 21st-century initiative from the FDA prescribed in 2004 that risk-based approaches and scientific use of information should be the basis on which to make decisions. In that context over a decade ago, ICH Q9 introduced the concept of risk management to the pharmaceutical industry and pointed out that QRM is a valuable component of an effective pharmaceutical quality system. In fact, QRM has played a significant role in the modernisation of GMPs (9) and more recently became mandatory and a de facto new GMP requirement in Europe with the publication in 2015 of Annex 15 to the EU GMPs Directives (10).
According to ICH Q9, three main steps are required to implement QRM: (1) risk assessment, comprising identification and analysis; (2) risk control; and (3) risk review. For successful implementation, important considerations include selection of the right QRM tools for correct identification of risks involved in the considered process.
Risk Identification
The step that is the main challenge involved in the QRM process is risk identification. Once the risks involved in a process are thoroughly identified, mapped, and understood in terms of causality, the next steps tend to be successful. In this work, reliable sources of information were consulted, that is, the WLs issued by the FDA and non-conformity reports (NCRs) issued by the European Medicines Agency (EMA) (11). These documents are an accurate source of valuable information regarding the identification of failures introduced by the manufacturer. The WLs and NCRs characterize the real situation in terms of quality and GMP compliance found across the pharmaceutical industry (7, 6).
QRM Tools
Several tools are available to support all the QRM phases, with different benefits and limitations. In this work, three of those tools were used: the Ishikawa (i.e., Fishbone) diagram, the preliminary hazard analysis (PHA), and the failure mode and effects analysis (FMEA) (8).
The Ishikawa diagram is a less formal and quantitative tool than other tools and allows the identification of possible causes of an event and organizes them into categories. It was useful to study the relation and interaction between a noted non-conformity (NC) and the resulting microbial contamination (12). The PHA is a formal tool usually applied in early stages of a QRM and is considered the ideal tool to start the study because it can be applied when the information available is limited. The PHA uses two factors—severity of failure and likelihood of it is occurrence—which multiplied generate the risk ranking index (RRI). FMEA expands on PHA as it considers how a particular failure mode can be controlled. The main difference between PHA and FMEA is that the latter takes into account the detectability of the causes and/or effects of a particular failure mode, while PHA assesses only the criticality of an effect based on its frequency of occurrence and severity. FMEA creates a risk priority number (RPN) for each failure mode as the product of the three risk dimensions above (13⇓–15). From the RPN profile risk ranking or prioritization enables efforts to be directed towards the most critical risks.
Methodology
Risk Identification
The WLs and NCRs are documents publicly available in the FDA and EMA databases, available from their respective websites. In this paper, we considered non-conformities issued by both authorities for manufacturers of non-sterile drug products between January 2008 and February 2016.
The FDA WLs database includes letters issued to different types of industries. In this work only those from the Center for Drug Evaluation and Research (CDER) were examined.
The NCRs were collected from the EudraGMDP database, which is maintained and operated by the EMA. In this work only those from the cGMP Department were examined. Figure 1 illustrates all steps carried out to obtain data from WLs and NCRs and subsequent steps used to evaluate the retrieved data.
Workflow used to analyze the retrieved warning letters.
Risk Analysis
Once the non-conformities were identified and organized according to cGMP violations, a risk assessment was conducted. After risk identification, a cause and effect analysis was performed through an Ishikawa diagram. The NCs found in WLs and NCRs were categorizedinto the six categories of an Ishikawa diagram: mother nature, manpower, machinery, materials, measurements, and methods. This classification allows mapping and subsequently understanding the NC root cause link to its effect (microbial contamination) (12).
Then PHA and FMEA were conducted with the cooperation of a panel of 14 subject matter experts (SMEs) from pharmaceutical manufacturers located in Portugal, within the framework of an ad hoc forum hosted by INFARMED (National Authority of Medicines and Health Products, Portugal), herein called the ISO Panel (16). The PHA and FMEA were generated based on the average classification given by the ISO Panel members for the three descriptors—severity of harm, likelihood of occurrence, and difficulty of detection—as follows:
Severity of Harm (PHA and FMEA)
High: the consequences of this failure mode are very important
Moderate: the consequences of this failure mode are moderate
Low: the consequences of this failure mode are low
Likelihood of Occurrence (PHA and FMEA)
High: the failure mode occurs frequently
Moderate: the failure mode occurs periodically
Low: the failure mode occurs rarely
Difficulty of Detection (FMEA)
High: the failure mode will be very difficult to detect
Moderate: the failure mode might be detected
Low: the failure mode will very likely be detected
To facilitate the discussions, results from PHA were assembled into three RRI categories (Table I): low (scores 1–3), moderate (scores 5–9), and high (scores 15–25); while the results from FMEA were grouped into three RPN categories (Table II): low (scores 1–9), moderate (scores 15–27), and high (scores 45–125).
Quantitative Evaluation of Risk
Quantitative Evaluation of Risk Using Three Dimensions Factors
Results and Discussion
Risk Analysis
From January 2008 to February 2016, the FDA issued 143 WLs, while the EMA issued 112 NCRs. Of these, 31% and 25%, respectively, were issued to manufacturers of non-sterile finished drug products. Regarding that subgroup, 52 cGMP violations in the WLs and 63 in the NCRs were identified.
The most frequent non-conformities by manufacturers of non-sterile drug products and their frequency of occurrence are listed in Table III. Note that some of them may have an indirect impact on microbial contamination. In this study, all NCs were considered and evaluated as to their potential impact and criticality. The ISO Panel was challenged to rank the NCs in terms of potential impact on microbial contamination and subsequently provide occurrence and detectability scores. This was done to align criteria and to avoid a possible microbial risk being neglected, as the concept of risk or microbial risk could be different between the authors and the industry focus group.
Most Common Non-conformities Committed by Manufacturers of Non-sterile Drug Products as Well as Their Likelihood of Occurrence, According to the FDA WLs and EMA NCRs
Ishikawa Diagram
Classifying each NC according to one of the six categories in an Ishikawa diagram made it possible to visualize the most frequent causes of contamination. Furthermore, this classification allows calculation of the percentage of NCs ranked in each category, by dividing them by the total of NCs found in WLs and NCRs (52 for FDA and 63 for EMA). This information is relevant to mitigation action plans aimed at controlling microbiological contaminations.
The diagrams produced with EMA and FDA NCs gave comparable results (Figure 2): materials (FDA 12%; EMA 10%), mother nature (FDA 1%; EMA 11%), machinery (FDA 12%; EMA 13%), manpower (FDA 27%; EMA 27%), measurements (FDA 15%; EMA 14%), methods (FDA 33%; EMA 25%).
Ishikawa diagram representing the cause and effect study conducted with NCs obtained in WLs issued by the FDA and EMA.
These results highlight two categories, “Methods” and “Manpower”, that together received half of NCs identified by the FDA and the EMA. This means that NCs related to the process performance and the specific requirements for the process, as well as NCs related to human error and education, were more commonly cited by inspectors. Indeed NCs grouped into “Methods” categories point out failures concerning, for example, lack of scientifically sound procedures, specifications not being defined, written testing programs, and “lack of control procedures”. Concerning “Manpower”, the problems identified were mostly related to failure to follow procedures and activities, demonstrating a lack of training and education concerning cGMPs and lack of evidence of a quality culture by employees.
PHA and Risk-ranking Index
The PHA analysis carried out in collaboration with the ISO Panel showed the highest microbiological risks involved in the manufacture of non-sterile drug products. The results obtained revealed that failure in quality-related issues represents the highest risk for microbiological contamination. Additionally, 50% of the FDA's NCs were considered low risk and 50% moderate risk. Regarding the EMA, 7% were considered high risk, 76% moderate risk, and 17% low risk. An overall assessment of the NCs that obtained the highest scores is presented below.
Moderate Risks:
Raw material quality control (FDA): Raw materials are reservoirs of microorganisms and are considered one of the most important sources of drug products contamination (17). Therefore, assuring their quality is a crucial step in avoiding microbiological contamination in the finished product. Before the implementation of a reduced testing regime, pharmaceutical industries should take into consideration supplier qualification, and the reliability of certificates of analysis should be checked regularly (18). That is, “Your firm has not established the reliability of the supplier's analyses through appropriate validation of the supplier's test results at appropriate intervals.”
Failure to establish and follow adequate control procedures (FDA): To avoid microbial contamination, pharmaceutical companies have the responsibility of monitoring and establishing process controls, especially those related to manufacturing personnel and the manufacturing environment. The manufacturing process should be evaluated for its potential to limit or eliminate bioburden (18). A microbiological control strategy should be planned by the manufacturer, and it should be based on a previous risk assessment. This task includes defining measures to limit microbial content and specifying microbial quality attributes (18). “Your firm failed to establish and follow appropriate written procedures, designed to prevent objectionable microorganisms in drug products not required to be sterile.”
Failure to assure the quality of the drug product (FDA): Batch release testing is only the final control step and does not provide sufficient assurance of product quality. Quality should, therefore, be evaluated and acquired during the process. However, industries should be able to test the finished product and to release a certificate of compliance for non-sterile drug products, which should comply with the specifications for total aerobic microbial count (TAMC) and the total combined yeasts/moulds count (TYMC). “Your firm does not have, for each batch of drug product, appropriate laboratory determination of satisfactory conformance to final specifications for the drug product, including the identity and strength of each active ingredient, prior to release (19).”
High Risks:
Failure to raise an Out-of-Specification (OOS), related Corrective Actions Preventive Actions (CAPA) investigations: GMPs require that an investigation be conducted whenever an OOS test result is obtained (20). The purpose of the investigation is to determine the root cause of the OOS result, to define the impact for patients, and to define corrective and preventive actions. Therefore, failure to take those steps and actions, especially related to microbiological OOS results, may lead to a release of a contaminated batch. That is, “Failure to raise OOS, related investigations and CAPA.”
Cleaning procedures: Cleaning is crucial to minimize bioburden, and its validation is essential to ensuring the efficacy and reproducibility of the cleaning process. Effective cleaning, disinfecting, and drying of equipment and facilities means a risk reduction relative to the proliferation of an opportunistic and adaptable microrganism. Microbial control of a cleaning validation program ensures that microbial residues are reduced to an acceptable level and are therefore under control. “Inappropriate validation of cleaning procedures.”
Quality agreements: Agreements between owners and contract facilities should define the cGMP-related roles and manufacturing operations. Therefore, the owners should ensure that the contract manufacturing facilities have a microbial contamination control program, and that such programs cover the drug product life cycle (20). It is necessary to review and evaluate the quality agreement at regular intervals to certify that this contamination control requirement is effectively in place. “Quality Agreements for production on behalf of Third Parties were not updated.”
The highest RRI scores from PHA show that failures in quality-related issues represent the highest risk for microbiological contamination (Table IV). Quality assurance failures are related to deficiencies in a company's quality culture (21).
Risk-Ranking Index and the Highest Scores Obtained
Another important issue identified in the RRI is related to raw material quality control, which is a crucial issue because it may have a direct impact on microbiological contamination of drug products. In fact, the recalls registered on FDA enforcement reports between January 2008 and February 2016 show that this was a frequent cause for the contamination of non-sterile drug products (22). Therefore, this NC may have been undervalued by the ISO Panel.
FMEA and Risk Priority Number (RPN)
Table V summarizes the RPN scores obtained through FMEA. As mentioned, the inclusion of “difficulty of detection” in FMEA measures the capacity of the pharmaceutical quality systems in place to detect an NC. As expected, the NCs that are more difficult detect represent the highest risk to the system and therefore need more urgent and close attention in terms of implementation of mitigation actions.
Risk Priority Number and the Highest Scores Obtained
Expected failures related to quality agreements are of moderate risk, according to our ISO Panel, but because they received the highest scores in difficulty of detection they must be analyzed as first priority. Quality agreements, especially on behalf of third parties, may have an indirect impact on microbial contamination because these establish responsibilities and assure that each party involved in the manufacturing of a drug product complies with cGMP regulations. Therefore, quality agreements should describe the roles and responsibilities of product owners and contracted facilities especially related to microbiological test methods, sampling, and microbial control strategy, as well as to manufacturing process controls (21).
Other important issues that should be prioritized are failures related to process controls. Process controls are crucial in avoiding any kind of contamination, especially those related to microbial analysis, manufacturing premises, good process design, and strict daily operational controls; all are key to robustness of performance as well as to effective microbiological control.
Risk Control
Once the risks are identified and quantified it is necessary to define strategies to deal with them. Several strategies can be adopted for avoidance, acceptance, transfer, and mitigation of high risks identified. According to ICH Q9, the amount of effort and attention for risk control should be proportional to the significance of the risk. The risk strategy should include benefit-cost analysis (13).
To ensure the appropriate risk acceptance and avoid neglecting some failures, it is important to set a level of acceptance. In this way, any risk with a severity score classified as moderate (score 3) or high (score 5) needs to be reduced as low as possible, irrespective of the RRI or RPN obtained. This level of tolerance is important because such scenarios could result in a significant impact to product, to process, or even to a patient.
For microbiological contaminations, the risk acceptance is very low, given the potential high impact on drug product quality and on a patient's health. To assess the quality of a non-sterile drug products and define which risks are acceptable, the manufacturer should perform a dedicated and detailed risk assessment exercise, adapted to their own reality, considering factors such as: (1) the intended use of the drug product, such as dosage form and route of administration; (2) the identity and specificities of microorganisms responsible for contamination; (3) the product characteristics, such as composition and formulation; and (4) the potential impact on the population receiving the drug product (23, 24).
Therefore, it is a good principle to analyze each NC individually, as they may have the same RRI or RPN but originate from very different combinations of severity, occurrence, and detection and may have different levels of acceptability (11).
Table VI gathers all the NCs that received the highest scores in RRI and RPN. Failures related to quality agreements, process control, investigation of deviations, and personal training were considered the most critical NCs. In fact, these findings reveal that quality systems are crucial in effective microbial control strategies and reinforce the need to establish a strong quality culture in the companies.
Summary of Results Obtained with FMEA and PHA for Non-conformities Identified by the FDA and EMA
Most non-conformities identified in WLs issued by FDA and NCRs by the EMA are related to materials and manpower. Therefore the first preventive action that should be applied is more effective and regular personnel training and awareness, together with more effective ways to perform documentation reviews. Each company should evaluate its own manufacturing scenario and be attentive to the types of failures experienced by peer companies, as descibed by the methodolgy used in the current work.
Conclusion
Regulatory agencies possess large amounts of information collected during inspections, of which WLs are a condensed and a formal summary, focused on significant GMP non-conformities, which are publically available in open databases. A laborious screening of such information allowed us to perform a cause and effect analysis, as well as develop a risk-based approach to describe the most likely failure modes that could be involved in microbial contaminations of non-sterile finished drug products, over a period of 7 years of WLs issued by the FDA and NCRs issued by the EMA.
The QRM tools used allowed a better understanding of the nature of noted NCs: through an Ishikawa diagram and its six categories it was possible to establish the likely causes for the most frequent NCs, and through PHA and FMEA a ranking was achieved according to likely criticality and frequency of those causes, so that mitigation strategies could be planned and applied. In summary, manufacturers of non-sterile drug products should focus their attention on the quality agreements, especially on behalf of third parties. In general, the quality agreements should be specific and clear about microbial control strategies to be carried out by each party. Other issues that are essential and deserve to be highlighted are the improvement of process and quality controls, especially those related to raw materials (including active pharmaceutical ingredients, water, excipients), containers and closures, and stability testing programs. A robust design of processes and facilities, together with strict daily operational controls, are crucial in avoiding microbiological contaminations. Relying only on finished product testing is not adequate for controlling microbiological contaminations.
Pharmaceutical companies should promote their own understanding about specific microbiological risks and use the data gathered in those studies to evaluate, for example, their own microbial ecology, their own manufacturing processes, and particular conditions and dosage form characteristics. For example, ICH Q6A provides decision trees to guide manufacturers on possible test strategies based on the nature of the product. In ICH Q6A, Decision Tree #6 for drug substances, and Decision Tree #8 for drug products, describe which microbiological attributes should be considered (18). A better and more robust industry, in terms of reduced risks of microbial contaminations, will emerge from the implementation of science-based approaches to risk management and enforcement of a quality culture that intrinsically reduces risk behaviors and attitudes.
Conflict of Interest Declaration
The authors declare that they have no competing interests.
Acknowledgments
The authors would like to thank the ISO Panel and INFARMED (National Authority of Medicines and Health Products of Portugal) for their participation, their support, and all their contributions to the risk assessment material presented here.
- © PDA, Inc. 2018
