Control Cross-Contamination Risk with HAPIs: Using a Risk-MaPP Master Matrix

Oct. 12, 2011
How Alkermes implemented Risk-MaPP to maximize operator and patient safety.

Recently, an international regulatory agency challenged Elan Drug Technologies’ manufacturing services business, now part of Alkermes plc, to develop an approach for managing risk at its multiproduct facility in Athlone, Ireland. The plant processes many different active pharmaceutical ingredients (APIs), including high potency APIs (HAPIs).

Working closely with PharmaConsult Ltd.’s Stephanie Wilkins, cochair of ISPE’s Risk-MaPP task team, the facility responded with a Master Matrix based on Risk-MaPP, a risk-based framework derived from principles outlined in ICH Q9. Risk-MaPP is designed to allow companies to identify and focus on critical risk areas to help prevent cross-contamination and ensure that controls applied are appropriate and commensurate to the risk. The Master Matrix, in turn, is a spreadsheet that offers a visually effective way to see where cross-contamination risks, both process-related and product-related, are highest, allowing appropriate actions to be taken.  This tool allows Alkermes to assign a numeric value, not only to each potential source of cross-contamination risk, but also to products that are vulnerable to cross contamination.This article will briefly summarize how the Master Matrix was implemented and what results it has shown so far.Alkermes considers the “opportunity to contaminate” as a function of the frequency of the given product’s manufacture. In addition, the company characterizes source batches or dosage forms at risk of cross-contamination by evaluating the following:•    toxicity •    quantity of active used per batch•    process train used in product manufacture •    level of containment and the energies employed in processing•    proximity to other products and the use of shared equipment •    opportunity to contaminate •    dosing regime of the product and in particular the number of daily doses contained in a batch•    frequency of the ingredient’s or product’s manufacture•    any other products manufactured on site that might be contra-indicated for users of the target drug.The goal is to identify and highlight points at which high risk and high vulnerability coexist.  The elements of the matrix can be divided into five subsections or “elements.”1. Product DataThe first step in building the Master Matrix was to list products manufactured on site.  Internally, to simplify communication, product names were used, while externally (and below) product numbers were employed to ensure confidentiality (See Table 1).
Populating the columnsFor each product, the applicable therapeutic area, dosage type, and route of administration are described. While Alkermes’ Ireland Facility is primarily a solid dose producer, liquids and a small number of injectable products are also manufactured. Clearly, information on the route of administration and on the targeted use of the drug product informs those reviewing the matrix as to how compromised the defences of patients using these products might be.  Keeping the patient in mind at all times, the matrix also notes the “contra-indication” of drugs.  While this has not proved as useful as originally expected, this was included to highlight any areas where a drug product may be manufactured adjacent to or sharing equipment with, another product whose patient information leaflet (PIL) recommends that they not be used by patients in combination.At the Athlone site, Alkermes also handles (largely in development) a small amount of controlled substances or scheduled drugs. While not of any special concern in cross-contamination assessments, a column to highlight such APIs has been included on the matrix.Numeric valuesThe first significant numeric value entered in the matrix is the Acceptable Daily Exposure (ADE)—a measure of an active ingredient’s toxicity defined simply as “the amount of active a person can take every day for the rest of their lives without it having an adverse effect.” For the purposes of Alkermes' Master Matrix, this measurement was interpreted as the maximum allowable daily amount of contaminated active present in one of its manufactured products to which a patient could be exposed without experiencing the therapeutic/adverse effect of the contaminator. Alkermes' view that the ADE should in all cases consider all potential sub-populations negates the need to highlight susceptible (contra-indicated) sub-populations in the matrix.  Of particular importance is that the ADE information is provided in a consistent and auditable way. As the ADE is a main pillar in the Alkermes matrix, good scientific assessment and judgment must be applied. Alkermes has based its approach on that described in Section 5.3 of the ISPE Risk-MaPP guidelines. Additional requirements—including the availability of key source data for audit review, qualification of toxicologists performing analysis, as well as checks performed and signed for by the author and reviewer—have also been added.  For non-toxicologists, reviewing ADE reports from a variety of sources has proved the greatest challenge, because understanding factors used in derivations reflecting “professional judgments,” “completeness of database,” “similarity to other molecules” and such, are difficult to standardize. In addition differences in judgement and terminology were also encountered. The solution was to request and require “peer review” of the reports involved, to build an independent audit review process on sampled reports (at least initially) and to ensure that providers used the terminology and factors listed in the ISPE guidelines. Furthermore, Risk-MaPP provides guidance on the quantification and measurement of various uncertainty factors, allowing definition of default values. Providers were required to justify when deviations from these default values were chosen. Alkermes used external contracted toxicologists for this body of work.2. Amount of API Needed to Contaminate ProductHaving conservatively established the ADE limits to include all potential sub-populations, it is next important to understand how much unintended active would be needed to contaminate a drug product in order to harm. Simply, it is necessary to relate the maximum potentially prescribed daily dose of all products in manufacture to the actual manufacturing batch size.Consider “x” grams of contaminating active getting into a blend. Assume that contaminant is evenly spread through the blend during processing. If the blend produces 1,000 (maximum) daily doses, then each will contain 0.1% of the contaminating material. If instead the blend produces 1,000,000 daily doses, then the patient will receive 0.0001% of the unwanted contaminating material with the prescribed drug product. Clearly the potential to exceed the ADE is far greater where the blend produces fewer “daily doses.”  The toxicity (ADE) of the potentially contaminating material and the amount of it that a patient might unknowingly consume are essential elements in generation of the risk profile (see Table 2).
3. Determination of Process Risk Proximity of processing is a major consideration in the estimation of risk. As a result, Alkermes' manufacturing areas were separated into zones based on processing rooms that shared a common corridor. These zones were then assigned a color and number to allow for easy identification. The process stages were then mapped for each product in these zones. (View the map in PDF format.)In the matrix (Table 3), the steps employed in each product’s manufacture (material number shown in left-hand column) are identified (across the top row) by coloring the cell at the intersection point of column and row. For each zone of Alkermes'  manufacturing facility, a different color is used. In this way it is immediately evident which products are manufactured in the same area(s) of the plant. 
The format outlined in Table 3 has since been refined. However, what is important to remember is that processes are not the same and are not equally likely to generate contamination. Factors including level of containment, mechanical energy input, ratio of product quantity to surface area of equipment, as well as the physical state of material (e.g., is it in solution, is it a coated bead, or is it a dry powder?), all have an effect on potential to contaminate. Alkermes uses a number of sources to grade the potential of individual processes to generate contaminants (process risk). For instance, assessment is generally informed by industrial hygiene (IH) data gathered during operator exposure testing.  Alkermes uses the scale in Table 4 to represent process risk.
Alkermes’ initial approach to scoring of processes is also shown in Table 3. Note that some boxes have scores greater than 20. This occurs for two reasons:
  • Everything is related to a finished batch size; where multiple components are combined in that final batch, multiplication of each individual component score by the number of components combined in a batch is undertaken. For example, coated beads are manufactured in the CF granulators in component batch sizes of 60-80 kg, but finished (encapsulation) batch sizes are bulked up by combining four of these in a blend. As a result, a 4x “component score” is entered in the matrix.
  • A number of Alkermes processes are iterative. Again, taking the example of our CF coated beads, typically multiple layers are used. The first is when active material is coated directly onto a non-pareil core. Such a process is “semi-open” with “medium to high” energy input and is scored a 12. The next coating is of a polymer onto the active core—the potential for active “release” exists but is much reduced as all active has adhered to the bead rather than existing in a powder state and auger feed to an agitated bed. Further, as polymer coating covers more and more of the bead surface, the product design virtually eliminates potential for contaminant generation. The second coat is scored a 10 and all subsequent coats as 5 or 2. All scores from the individual iterations are then combined.
The sum of the scores for each of the individual process steps yields an overall Process Risk score for that specific product. How this score is used to determine Product Risk & Vulnerability will be explained later and will give a better understanding of objectives and methodology employed. 4. Determination of Product RiskBy adding all steps used to manufacture a product, an overall process score or “Process Risk” is derived. This number is reflective of process design, of the level of mechanical energy input and of containment. It does not, however, consider the potency or toxicity of the API being processed, the amount of API handled or the opportunity to contaminate other products.The potency of a batch can be described by:(API/batch) ÷ ADE Simply, the more active ingredient used and the higher its potency, the greater the potential risk it presents to other products being manufactured. The more of a product made, the greater the opportunity or likelihood that contamination will occur. For example, if product X is made once a year using a defined process, there is less opportunity for elements to go wrong and allow it contaminate product Y than if the same product was manufactured using the same process five times a day.Combining these two elements with “Process Risk” provides a measure of the potential for the manufacture of a product to contaminate others, or “Product Risk.”Product Risk = (# batches manufacture/year) x [(API/batch) ÷ ADE)] x Process Risk
Calculating “Product Risk” for all products manufactured, allows them to be ranked and process trains presenting the greatest potential contaminating risk to others are immediately highlighted (Figure 1).5. Determination of Product VulnerabilityConsideration must now be given to the potential for a product to be become contaminated, described as “Product Vulnerability.”If a patient takes less than the ADE of a drug (assuming it has been calculated for all relevant modes of administration and to include all sub-populations), even if the patient takes it over an extended period, one can assume that there will be no effect.  Quantities prescribed to patients differ, so clearly the patient most at risk is the patient prescribed the largest envisaged daily dose. As remarked above, the smaller the batch, the less the dilution of contaminant and the greater its concentration in the finished product. These two considerations were combined (with opportunity) to rank the “vulnerability” of products manufactured.Vulnerability = (# batches manufacture/year) ÷ (# “maximum-daily-dose”/batch)Again, the opportunity to be contaminated is a function of the number of times a product is manufactured. See Figure 2.
Lessons LearnedTable 5 highlights that the greatest potential source of contaminant currently manufactured on the Alkermes Athlone Ireland site is Product “3000011.” The product most vulnerable to contamination, due to its manufacturing scale, frequency of production, and dosing regime is Product “3000014.”
Neither of these products are low ADE products and both have been manufactured for many years, yet the  analysis indicates that there should be a focus on where these products are manufactured and where they may come in contact with each other. In short, particular attention should be focused on  anywhere a product presenting a relatively high “product risk” is manufactured in close proximity to a product of relatively high “vulnerability.”. This is as described in the flowchart in Figure 3 and for Alkermes has meant that a detailed FMEA analysis is applied to any area where potential cross-over occurs.
Refinement of ApproachWhile developing the matrix to the point described above has been very useful and informative, limitations have been observed.1. The matrix needs to be regularly updated to reflect changing manufacturing volumes.2. Overall site scores do not immediately highlight potential issues in a shared manufacturing zone or corridor.3. The matrix does not provide an immediate understanding of a“vulnerable” process—e.g., if the product at risk of contamination is itself highly contained, whether the level of risk is reduced.4. The matrix does not differentiate short-run products (e.g., clinical or registration batch manufacture) where timing and lack of familiarity would heighten concern.Addressing the first of these issues allows for the relatively easy resolution of the other three and provides (in addition to a site picture) a very local assessment of cross-contamination potential.Because “vulnerability” does not include a measure of process containment, a “process risk” rating has been included in the revised spreadsheet. While not solely a reflection of containment, a low process risk score is generally indicative of a closed process, meaning that the path for contaminant is more problematic.
Figure 4. Graphical Representation of Equipment Process RiskThe revised sequence of matrix review now undertaken is:1. Identify highest “product-risk” and highest “vulnerability” products manufactured on site to provide an immediate high level focus on most important areas.2. Track products through zones of manufacture at a component level in the body of the matrix.3. Where high-risk sources coexist with the manufacture of vulnerable products, target detailed risk assessment to develop or ensure the adequacy of already in-place measures.
Risks have been submitted for each process stage. They are then added together to indicate where targeted improvement of process design might yield greatest impact. Although the Matrix is still a work in progress, Alkermes staff are well informed of considerations required for the safe design and operation of facilities to minimize cross-contamination. In areas of shared activity, review has prompted research and experimental verification. For example, the company has initiated a series of experiments to evaluate the protection afforded by air change rates, pressure differentials and physical barriers where two products are manufactured in processing rooms linked by a shared corridor.ConclusionProcedurally the matrix is formally updated annually or where a change (e.g., new product introduction) is deemed likely to significantly affect the plant’s risk profile.  Assessment of the impact of change is ensured through linkage with the site change control process, where now consideration of impact on cross-contamination risk is prompted not just by major product introductions, but also by equipment moves.There remains much to be done on education, on calibration with others in industry, on honing and improving Alkermes' process and on linked themes including setting of cleaning limits, but the Master Matrix has provided Alkermes a very strong foundation and framework on which to build.

About the Authors
Mark O’Reilly joined Alkermes (formerly Elan Drug Technologies) in 1993.  He is currently Senior Director of Engineering where his responsibilities include process train and facility design, industrial hygiene and validation. He received his BE from University College Dublin, his MEngSc from the University of New South Wales and his MSc from the University of Manchester.

Aisling Horan joined Alkermes (formerly Elan Drug Technologies) in 2003. She currently works in the validation department with a primary focus on process & cleaning validation and cross contamination. In addition to validation, her experience includes industrial and clinical microbiology. Aisling received her BSc (Hons) in microbiology from the National University of Ireland, Galway (NUIG).

About the Author

Mark O’Reilly and Aisling Horan | Alkermes