Application of Quality Risk Management To Set Viable Environmental Monitoring Frequencies in Biotechnology Processing and Support Areas

Quality Risk Management

Risk assessment and risk management have been associated, in some form, with pharmaceutical processing from the beginning, in line with the objective to produce a safe and efficacious medicinal product. In the 21^st century, however, risk management has been applied to the pharmaceutical industry in a more systematic and formal way (described as quality risk management). Here the application of risk management has been to focus resources on those areas that are the most important, to ensure science-based decision making, and to enable the full utilization of the resulting advantages (7).

Quality risk management has been promoted by European medicines inspectors and by the U.S. Food and Drug Administration (FDA) as central to 21^st century GMPs. The first major approach was by the FDA in the FDA document “Pharmaceutical cGMPs for the 21^st Century—A Risk Based Approach” (issued in 2004) (8). From this the widely used GMP approach to quality systems, outlined in ICH Q10 “Pharmaceutical Quality System,” followed (9), and this became adopted by EU GMP as Annex 15 to the EU GMP Guide (10).

The ICH documents outline quality risk management as a systematic process for the assessment, control, communication, and review of risks applicable to the quality of the medicinal product (11). One of the documents that forms part of the ICH quality set is the ICH Guideline titled Quality Risk Management (ICH Q9) (12). This document is of relevance to this paper, as it outlines a series of recommended approaches for risk assessment and risk review. The two overarching principles for quality risk management that ICH Q9 sets out are that evaluations of risk should be scientifically based and ultimately linked to risk to the patient.

Within ICH Q9, the major risk analysis tools described include FMEA (failure mode and effects analysis), FTA (fault tree analysis), and HACCP (hazard analysis critical control points); although there are other approaches (13), these tools are particularly useful in helping to deconstruct the complexity of pharmaceutical operations.

The Risk Assessment Process

Risk assessment is concerned with identifying and evaluating risks. Risks are generally recognized as being related to a situation, event, or scenario in which a recognized hazard may result in harm. Hazard, in this context, refers to any circumstance in the production, control, and distribution of a (pharmaceutical) product that can cause an adverse health effect. This often refers to a biological, chemical, or physical agent. Hazards and risks mean different things. A hazard is an ever-present property. When hazard and vulnerability interact together, this creates risk, and when something occurs, this is often described as an incident (with pharmaceutical processing, this could be an out-of-specification result or microbiological data deviation). Risk is therefore something that has a dynamic property, that is, it can alter over time according to different factors or circumstances (14).

Most risk assessment activities involve the quantitative assessment of risk based upon the magnitude or severity of the risk and the probability that the risk will occur (15). ICH Q9 defines risk as the “combination of the probability of occurrence of harm and the severity of that harm.” Of these concepts, probability is arguably the most important concept because what matters is whether something is likely to happen. Probability can be assessed from a combination of frequency-based calculation and a “degree of belief” (16).

Three key definitions are outlined in ICH Q9 that help to contextualize what is meant by risks:

• Risk: “The combination of the probability of occurrence of harm and the severity of that harm”

• Harm: “Damage to health, including the damage that can occur from loss of product quality or availability”

• Hazard: “The potential source of harm”

Risk assessment involves identifying risk scenarios either prospectively or retrospectively (17). With the former, this involves determining what can go wrong in the system and all the associated consequences and likelihoods; with the latter, this looks at the impact of what has gone wrong by assessing the process, product, or environmental risk and formulating the appropriate actions to prevent the incident from re-occurring (18). This paper addresses prospective risks.

Environmental Monitoring

Environmental monitoring refers to the microbiological testing undertaken in order to detect changing trends in microbial count and types of microflora within cleanroom environments. Individual results (whether of a high count or a low count) are, when considered separately, rarely significant, whereas problematic trends may indicate a risk to the products or processes undertaken within the cleanrooms. The results from environmental monitoring obtained provide information about the performance of the physical design of the room (most notably air handling systems) and the performance of the people, equipment, and cleaning operations within the cleanroom (19).

There are many elements that make up a microbiological environmental monitoring program. With these elements, some regulatory guidance is provided in relation to sterile manufacturing, particularly aseptic processing. However, for the early-stage processing of sterile products and for biotechnological manufacturing, guidance is more limited and there is no agreed best practice consensus (and arguably, given the variety of approaches and processes, such a consensus may not be achievable or desirable).

As a minimum, an environmental monitoring program should address the following elements (20):

• Selection of sample sites

• Maps showing sample locations

• Sampling procedure

• Sampling frequency

• Processing samples

• Sample incubation

• Data analysis, including trending

• Investigative responses to exceeded action levels

• Statistical data trending

• Establishment of alert and action levels

• Level of product exposure

• Consideration of the risk associated with formulation, process step, and dosage form, particularly with reference to undesirable or objectionable microorganisms

• Responsibilities

These elements should be captured in a formal program with a supporting rationale. Of the above list, the question considered within this paper is the sampling frequency. With a large suite of cleanrooms, how often should each be monitored?

Application of Risk Assessment and Risk Filtering To Determine Monitoring Frequencies

To illustrate the risk management approach this paper presents a case study. The objective was to study cleanrooms used for biotechnological processing and to assign risk-based monitoring frequencies from a review the design of cleanrooms and their use.

Case Study

To illustrate the risk assessment approach for assigning frequencies of viable environmental monitoring, the paper draws upon a case study. The case study was a manufacturing facility. The facility consisted of a suite of EU GMP Grade C (equivalent to ISO class 7 in the “at rest” state, in relation to airborne particle classification) and EU GMP Grade D cleanrooms (equivalent to ISO class 8 in the “at rest” state) used for intermediate product processing. In total there were 50 rooms at EU GMP Grade C and 40 rooms at EU GMP Grade D.

The monitoring frequencies considered prior to the start of the risk assessment were termed AVMF, an abbreviation for “assigned viable monitoring frequency.” The objective of the exercise was to assign a risk-based AVMF to each of the cleanrooms. The selection of appropriate monitoring frequencies, in the absence of regulatory guidance, is a subjective process based upon what appears reasonable, what fits in with available resources, what relates to the capabilities of the area, and what is consistent with the environmental monitoring performance. For the case study, monitoring frequencies were based, after review, from the commissioning data. It is possible, after a period time, that such frequency categories can be reviewed. With the case study the frequencies used are shown in Table I.

With the above frequencies, different patterns will be appropriate to different types of facilities. The process of linking each cleanroom with an AVMF was through a step-by-step approach, which involved

• a) Identifying the main risk factors common across each of the cleanrooms

• b) Evaluating each risk factor for the severity of the risk and the probability that the risk would occur

• c) Assigning a numerical score, based on a range of 0 to 4, to each risk factor for either severity or probability or for both, depending upon the nature of the risk

• d) Relating the severity and probability scores to each other

• e) Considering the detection methods in place

• f) Using a risk filter to assign a cleanroom to the required monitoring frequency

This step-by-step approach is discussed in the section below. The approach was within the risk management guidance outlined within ICH Q9, in that

• An evaluation of multiple diverse quantitative and qualitative factors for each risk was undertaken

• The basic risk question was broken down into as many components as needed to capture factors involved in the risk

• The factors were combined into a single relative risk score that can then be compared, prioritized, and ranked.

Risk Assessment Methodology

The risk assessment approach consisted of identifying the primary hazards, common to different cleanrooms, that would affect the cleanliness or microbial levels within the cleanroom environment. Of these, some hazards would pose a direct risk to the product being processed, whereas other hazards pose more of an indirect risk. This process was undertaken by assembling a multidisciplinary team, for it was important that there was an in-depth understanding of the cleanrooms and processes, to assist with the collection and review of the information.

Hazard Identification

For the identification of hazards, a HACCP (21) technique was used. The use of HACCP directed the team towards identifying those parameters that affected product safety or quality (22).

The parameters identified were

• Temperature

• Wet or dry areas

• Floor drains

• Processing activity

• Duration of activity

• Cleaning and disinfection frequencies

• Location of the process in relation to final formulation

• Room occupancy

• Fixed and mobile equipment

• Environmental monitoring history

Once each parameter had been identified, it needed to be assessed. This involved weighing how significant the risk was. This was undertaken by assigning a score range to each hazard for either the potential severity of the hazard or to the probability of its occurrence. For this exercise, a score range from 0 to 4 was adopted. As with any risk assessment tool, it is important to agree and to establish a scoring system prior to starting the exercise. The preference with this case study was not to have a score range that was too wide (such as 0 to 10) due to the over-complications that this would generate (such as debating the difference between, say, a score of 5 and 7).

In terms of examining each parameter for severity and probability, severity was defined as the considered impact upon the product. This was assessed through the use of three categories:

• Unlikely to cause product contamination (low severity)

• Low possibility of product contamination (medium severity)

• High probability of product contamination (high severity)

Probability was defined as the likelihood that the hazard could lead to the area becoming contaminated. This was similarly assessed through the use of three categories:

• Rare contamination events (low probability)

• Infrequent contamination events (medium probability)

• Frequent contamination events (high probability)

When setting severity and probability scores, a degree of subjectivity was inevitable. Concerns in relation to bias were counterbalanced through the use of a multidisciplinary team to agree on the assigned score. The division of severity and probability into low, medium, and high categories was necessary for the risk filtering process (as outlined in Table XII below). From this conceptual starting point, each hazard was assessed. For the assessment each hazard was broken down into its constituent parts, or sub-factors. The identified hazards are discussed below.

• a) Temperature This factor was assessed based on the typical temperature in each cleanroom. Here the areas within the manufacturing facility were grouped into freezers, cold rooms, ambient areas, and warm areas (such as areas where the temperature would sometimes exceed 30 °C, such as where heat sterilizers are unloaded). The areas were evaluated in terms of severity (how serious would microbial contamination be should it occur?) and scored. The score range this analysis is shown in Table II.

• b) Wet or dry state Whether areas were wet or dry was considered in terms of probability. This was in consideration of whether the state of cleanroom was likely to encourage the growth of microbial contamination. The factor was assessed by determining if the cleanroom had a primary water source (such as fitted sinks or if the area was a wash-bay). The scoring system outcome is displayed in Table III.

• c) Drains Whether or not the cleanroom contained a floor drain was considered separately to the wet/dry state. Drains can generally be considered as a risk mitigator because the absence of drains may result in more water on the floor. However, for improperly designed drains there can be a risk due to the potential for backflow or the formation of a biofilm. In this case study, based on swab results, the presence of drains was considered to be a potentially high risk compared to dry areas with no drains present. This hazard was considered in terms of probability (a drain is either present or not) and scored accordingly, as shown in Table IV.

• d) Processing The risk factor relating to processing considered whether open or closed processing took place within the cleanroom, or if the room was a nonprocessing area (such as a chemical store). The risk was weighted towards open processing because the product is exposed to the environment for a period of time. This was considered in terms of the severity of the risk, as displayed in Table V.

• e) Duration of activity In relation to processing, the duration of the activity for the processing of the product was considered. For this, arbitrary durations were selected based on the assumption that the longer the activity takes place within the room, then the greater the likelihood that contamination could occur and the more severe the consequences could become. Therefore, the scoring system considered both severity and probability as indicated in Table VI.

• f) Cleaning and disinfection With the frequencies of cleaning and disinfection, this was based on the premise that the more frequent the scheduled cleaning and disinfection, then the lower the probability of contamination build-up. There is a risk in evaluating this factor that a higher level of cleaning may have been implemented due to an upward trend with environmental monitoring, thus the factor considered the scheduled cleaning regime only. The scoring outcome is shown in Table VII.

• g) Location of the process With location of the process, the distance of the processing step within the cleanroom from the point of the final formulation of the product was considered. This was based on the concept that the nearer the process was to the finished product, then the greater the risk any contamination presented. This was because any steps that might remove or reduce contamination, such as dilution, the addition of chemicals, or processes like ultra-filtration, were less likely to occur towards the end of the process. The outcome of this assessment of severity is displayed in Table VIII.

• h) Room occupancy With the room occupancy rate, given that personnel are the largest source of contamination within cleanrooms (23), this factor was weighted towards the more personnel within the room leading to greater the chance of contamination being present. The relationship between numbers of personnel and different levels of contamination can be difficult to determine, although some historical evidence can be gathered from a review of environmental monitoring data. The outcome of this probability is shown in Table IX.

• i) Equipment With the equipment located within the cleanroom, there was considered to be a greater risk for equipment taken into and out of the cleanroom compared with equipment which remains fixed within the room. This was based on the premise that most fixed equipment is connected to fixed cleaning systems such as clean-in-place (CIP) or steam-in-place (SIP) mechanisms. Mobile equipment, in this case study, was more often manually cleaned and transported into and out of the cleanroom, which could present a greater chance of contamination occurring. It is noted that other facilities may have CIP stations to which mobile equipment can be taken. This assessment is outlined in Table X.

• j) Environmental monitoring The final hazard considered was the environmental monitoring history of the cleanroom. This was based on the probability that the more frequent the level of excursions above the action level, then the greater chance there was of product contamination. The criteria for what constitutes good or bad environmental monitoring will vary between facilities, although such an assessment will be based upon data trends. The assessment for the case study is displayed in Table XI.

Risk Filtering

In order to assess each of the hazards, the concepts of risk ranking and risk filtering were used to structure the collected the information and to determine the relationship between probability and severity, and then for relating these risk factors to the environmental monitoring frequencies.

The relationship between probability and severity was expressed through the use of a risk matrix to enable risk ranking (Table XII). Risk ranking is defined as the sum of the Severity and Probability rankings. From the assessment of the hazards and taking account of the scoring system, the possible range of scores for the severity category was 0 to 14 and the possible range of scores for the probability category was 0 to 16.

Based on there being three different categories for severity and probability, as outlined above, the score ranges were

The relationship between severity and probability, as a risk matrix, is shown in Table XII.

To the risk ranking arrangement above, three risk classes were assigned. These were

• Class One—high risk

• Class Two—medium risk

• Class Three—low risk

These risk classes were imposed upon the risk ranking matrix, as shown in Table XIII.

Monitoring

With the relationship between severity and probability established, the detection methods were considered next. The monitoring aspect is a fundamental part of risk management (24). Risk management approaches recommend that, following the identification of each hazard, an attempt should be made to eliminate the hazard or, failing that, to reduce the risk. When this exercise has been completed, the residual risk should be monitored to alert personnel to the risk becoming more severe or occurring with increased regularity.

In considering monitoring in relation to cleanrooms, the premise was that the more systems in place to detect cleanroom or product contamination then the stronger the detection system was. For the case study, three different types of monitoring were considered that would be applicable to each of the cleanrooms were used. These were

• Viable environmental monitoring

• In-process bioburden samples

• Process water samples

These aspects of monitoring were selected because the environment may potentially have some impact upon the product in relation to cross-contamination. With process water, this was used, in most cases, to formulate the product. In the case of water in wash-bays, the nature of the process led to periods of time when water is deposited onto the floor and onto equipment, thus the quality of the process water provided some additional, if indirect, assessment of the environment (this may not be applicable to all facilities). The in-process bioburden samples provided a direct measurement of the quality of material at a particular stage of processing and any contaminated could, through microbial identification, be cross-matches to the viable environmental monitoring results.

For each of the cleanrooms, at least one of the types of monitoring would be applicable (because, as a minimum, viable environmental monitoring could be performed in each room). For some rooms, two types of monitoring would be applicable: either an in-process sample could be taken to assess product bioburden or a process water outlet could be sampled (from water used for product formulation or for room or equipment cleaning). For a fewer number of rooms, all three types of monitoring were applicable.

For consideration of the types of monitoring, a scoring system was devised as shown in Table XIV, with the best detection system being the availability of all three types of monitoring.

Risk Filtering

Having established the detection methods (assigned as low, medium, and high), a risk filtering matrix was deployed. For this a second matrix was constructed where the detection methods were placed on the horizontal axis and the risk classes, drawn from Table XIII, were placed onto the vertical axis. From this arrangement, a relationship between risk class and detection could be discerned. With this arrangement different priorities for monitoring emerged through the detection method acting as a risk mitigation measure. This led to not all of the risk class one cleanrooms being considered as high-priority areas provided that a good detection system was in place (that is where all three types of monitoring were performed).

The resulting risk filtering matrix is displayed in Table XV.

In order to relate priorities for monitoring to environmental monitoring frequencies, the priorities were switched to the AVMF frequency of monitoring categories assigned earlier (displayed in Table I). This altered the risk filter matrix to read as shown in Table XVI.

Thus the high-priority areas became AVMF category 1 (weekly monitoring), the medium-priority areas became AVMF category 2 (biweekly monitoring), and the low-priority areas translated to AVMF category 3 (monthly monitoring).

Worked Examples

The discussion thus far has been largely abstract. To illustrate different outcomes more clearly, three cleanrooms, located in different parts of the facility and which were used for different process activities, have been drawn from the case study and are discussed below.

Review of Methodology

There are advantages and disadvantages with the application of risk assessment tools for the establishment of environmental monitoring frequencies. The advantages can be summarized as the following:

1. 1. Risk assessment provides a structured decision making method and allows the decision making process to be consistently applied.

2. 2. The scientific and data driven nature of risk assessment methodologies significantly reduces subjectivity.

3. 3. The risk management approach acts as a diagnostic tool in that it provides a framework to better understand processes.

4. 4. The structured approach can be presented in a form that can be shown to regulators in a form which is straightforward to explain.

5. 5. The philosophy of risk management is normally applied to many parts of organization, thus the environmental monitoring program can be integrated with other activities relating to pharmaceutical processing, such as validation, design, change control, and corrective actions.

6. 6. The approach allows the monitoring activities to be prioritized towards areas of greatest contamination risk. This approach additionally helps with effective resource allocation.

7. 7. Risk management promotes risk communication in that microbiology and production staff can each understand the relative risks to the environment and product from different process areas and activities.

There are, however, some disadvantages with the risk management approach for environmental monitoring. The first is that the selected frequencies may start off too infrequent and by the time sufficient data has been collected a contamination event may have affected the product or elevated the risk. In these circumstances it is easier to adapt the risk assessment approach to an existing facility, where there is sufficient knowledge of environmental controls, than within a newly built facility or recently modified process. With new facilities the standard approach is to subject these to intense initial monitoring during commissioning and then to set appropriate monitoring frequencies after a sufficient amount of data has been collected over a suitable time period.

The second disadvantage is that the weightings applied at the outset may be inappropriate, such as giving too much importance to a factor that does not matter greatly or too little importance to something that does matter. This can only be addressed through experience and from an understanding of the actual facility. For example, personnel occupancy may have a bearing on a room of a smaller size with air-change rates of 20 per hour but less so with a far larger room with air-change rates of 40 per hour. Each facility's HVAC (heating, ventilation, and air conditioning) system will be different and will affect the relative importance of different risk factors.

The third disadvantage is that changes to processes may take place that are not captured or only realized long after the event. Here, a robust change control process should capture changes to room use or design, or to process changes, and the information should be communicated back to those tasked with undertaking the risk assessment. The fourth disadvantage is common to all risk assessment approaches. Here, while approaches can be rationalized to a degree and scientific principles are applied, many of the risk methods adopted are subjective and rely, to an extent, upon supposition. Often with a contamination event there can be multiple risks and some of the risk assessment approaches are weaker when dealing with multiple risk situations than others. This is overcome, to an extent, by risk weighting and risk filtering.

A fourth disadvantage relates to a running concern with risk assessments. That is that is less experienced individuals may favor the use of risk assessment tools but would be unable to evaluate the results against their experience. This can be overcome by ensuring that a team is used to evaluate the results and that the team includes someone familiar with microbiology and risk assessments.

With these advantages and disadvantages acknowledged, the use of risk assessment as applied to environmental monitoring provides more benefits than problems and can lead to the construction of a program which seeks to monitor contamination more frequently from those areas that are at greater risk or that have weaker detection systems in place.