Abstract
Pharmacopeias recognize particulate matter as a common phenomenon. The current regulatory requirements relating to particulate matter in parenterals state that solutions for injections or infusions are clear and “practically” (or “essentially”) free from (“readily detectable”) particles when examined under defined conditions of illumination. Pharmaceutical companies are required to know their processes and have them under control. In order to control and reduce the potential influx of particulate matter, Novartis Technical Operations in Unterach developed a particle life-cycle program that involved an establishment of a dedicated particle laboratory operating under clean room conditions. The analytical capabilities of this particle laboratory were crucial for the characterization of particles and supported identification of potential sources of particulate matter. After implementing this program and respective actions, product and process understanding significantly improved. This resulted in a decrease of reject rates, AQL (Acceptable Quality Limit) failures, and corresponding batch rejections, thereby increasing the availability of Novartis products from Unterach for the patients. The main objective of this article is to show the detailed particle characterization approach including Quality by Design (QbD), methods, and equipment. Examples from projects and particulate matter investigations are presented.
LAY ABSTRACT: Parenteral formulations should not contain particulate matter. However, as external contamination as well as formation of particles during manufacturing cannot be entirely excluded, pharmaceutical companies use visual inspection and AQLs to monitor occurrence of particles. To ensure patient safety, Novartis Technical Operation in Unterach established a particle-free analytical laboratory with a focus on particle characterization and root cause analysis of particle formation. The ultimate goal is to reduce occurrence of particles in formulations altogether, and increase process understanding. The approach toward particle characterization adopted at Novartis Technical Operations in Unterach is presented in the article.
- Particle laboratory
- Quality by Design
- Microscopic techniques
- Spectroscopic techniques
- Process understanding
- Root cause analysis
Introduction
Among the various reasons for drug recalls, presence of particulate matter has played a prominent role (1). In the U.S., out of 45 drug recalls in the period April 2015 to March 2016, 23 involved injectable drugs. The recalls due to particulate matter were limited to only injectable drugs and caused 61% of their recalls (31% of all recalls), followed by sterility with 22% (see Figure 1) (2). Particle analysis is essential for continuous process improvements and life-cycle management of marketed products, but these are equally important in the development phase to provide better understanding of the formulation behavior. This is in accordance with the Quality by Design (QbD) approach, aiming to minimize and ideally eliminate risks of particle formation altogether. The focal point of both regulatory and industry representatives is to ensure patient safety.
Number of recalls and distribution of 23 recalls of injectable drugs in the period from April 2015 until March 2016 in the U.S.
The selected pharmacopeias' requirements relating to particles in injectables are listed in Table 1. Phrasing “essentially,” “practically,” or “readily detectable” appreciates the fact that an entirely particle-free product or environment in reality does not exist (3). Particles in the submicron range (<1 μm) are likely to be present and rely on the sensitivity of analytical methods to determine their count, if they can be detected at all. Recent recalls due to particles, as shown in Figure 1, highlight increased attention to the particle topic (2). Langille (4) and Doessegger et al. (5) discuss the clinical effects of injected particulate matter on patients. Particle size, shape, route of administration, count, composition, and patient safety are important factors to be considered when evaluating harm caused by particulate matter. Owing to the lack of clinical studies in humans, the most current knowledge is based on animal models. This lack of understanding of what risks the presence of particles in injectables pose to patients was highlighted by industry representatives (6). The industry, however, adopts a conservative approach to limit particles in formulations to essentially or practically free.
Pharmacopoeial Requirements for Visible Particles
Categorization of Particles and Scope of the Article
Particles are categorized based on either size as visible and subvisible or source as extrinsic, intrinsic, or inherent (Table 2) (10). A typical trigger for an investigation is an unexpected occurrence of visible particles. This article focuses on analytical techniques for characterization of visible particles in formulations based on small molecules, although the methods used for protein therapeutics are essentially the same. However, the analysis of protein solutions does have some specific peculiarities, which have been previously summarized in several reviews (11). Methods and requirements implemented to detect particles during the routine visual inspection of filled products are out of the scope of this discussion. Earlier publications covering general inspection techniques, influence of primary packaging material, particle physics, and methods of measurement were published by Borchert et al. (12) and Ernerot (13). A review of Borchert's study also outlines a scientific approach for analysis of particulate matter, and despite >30 years since its publication and improved instrumentation, the methodologies and though processes applied remain relevant even today. A classic resource has been McCrone Atlas of Microscopic Particles, available online (14). A detailed investigation scheme for particle analysis was also described by Aldrich and Smith (15). Control of particulate matter in manufacturing was discussed in detail by Barber (16).
Types of Particles
In this article, we aimed to provide an update regarding some of the methods that can be used for the identification of detected particles. Furthermore, we share our experience and describe our approach to handling and solving particle-related problems. We also want to highlight how the particle analysis relates to the development QbD approach. This article focuses on the analysis of particles present in liquid formulation in vials and does not intend to provide a detailed technical discussion of the methods. Although all investigations have been performed in a GMP lab, this article does not intend to discuss GMP-compliance issues. Of the particles listed in Table 2, inherent particles related to the active pharmaceutical ingredient (API) are arguably the most challenging to identify. The reason is that no reference spectra are often available for these API-related particles. Sometimes these particles have limited stability and are present in small amounts. Identification of extrinsic and intrinsic particles might be less difficult, but finding their origin and the root cause of their occurrence remains challenging.
Experimental Details
Particle-related investigations are ideally handled in an on-site dedicated particle laboratory, situated directly in the production area. At Novartis Technical Operations in Unterach, the particle laboratory is operating under cleanroom grade C conditions according to the EU GMP guideline. Manipulations with open containers are performed in safety workbenches under cleanroom grade A conditions, monitored by the building maintenance system. The EU GMP cleanroom particle limits (17) and comparison to the ISO standards (18) are shown in Table 3.
EU GMP Particle Requirements of Clean Room Grades A–C and Comparison to ISO Standards
We typically place the particles on gold-coated filters (pore size, 5μm; diameter, 25 mm) from Unchained Labs — rapID. A wide-zoom stereomicroscope (Leica M205A) and a compound light microscope (Leica DM2700M) are used to define optical properties of the particles (isolated or in the vial). Particles are photo-documented using the built-in camera systems of the microscopes.
Raman and laser-induced breakdown spectroscopy (LIBS) spectra are obtained using a Single Particle Explorer from Unchained Labs — rapID, model raman.ID and metal.ID. The gold-coated filter is fixed to a sample holder and placed into the instrument. In addition to Raman and LIBS, the Single Particle Explorer also contains a built-in microscope that is used to recognize and measure the size of the obtained particles and identify these prior to obtaining Raman or LIBS spectra. The microscope of the Single Particle Explorer can also be used to define the optical properties of particles.
Infrared spectra are obtained using Fourier transform infrared (FT-IR) microscope from Agilent (model Cary 610 microscope coupled with Cary 660 spectrometer).
The spectroscopic characterization (Raman, LIBS, and FT-IR) of particles is performed in general manually by comparing the obtained sample spectra with the reference spectra in the respective databases. We use Raman and LIBS spectral databases from Unchained Labs — rapID and FT-IR database from Agilent. We continuously expand all databases with internal entries.
The instruments listed above are routinely used in the particle laboratory in Unterach. However, instruments from other suppliers with similar or equivalent specifications can be used for analysis of particles.
QbD and Iterative Learning of the Particle Analysis
QbD adopts a proactive approach to address risks by the systematic analysis of components within the finished dosage form (FDF) and production steps on the product's critical quality attributes (CQAs) (19).
Particulate matter is, per definition, a CQA and has been evaluated in the quality risk management (QRM) documents at Novartis Technical Operation in Unterach from the beginning of the pharmaceutical development process.
Prior experience (relying on strong knowledge management within the organization), expert knowledge, and published literature help to assess risks, which are subsequently either disregarded or explored experimentally by development groups. However, each new product is, by definition, unique and it is possible that factors that were unanticipated, underestimated, or overlooked at the outset will emerge at a later stage of pharmaceutical development. Novartis Technical Operation in Unterach mitigates this risk by regularly updating the QRM documents.
We wish to emphasize the iterative learning process, where by “learning is achieved, not by simple theoretical speculation on the one hand, nor by the undirected accumulation of practical facts on the other, but rather by a motivated iteration between theory and practice” (20). Particle analysis at Novartis Technical Operations in Unterach views this learning cycle as the cornerstone of development and continuous improvement activities, with focus on the successful identification and resolution of specific particle-related topics. In the spirit of QbD, the potential formation of particulate matter should be included in the formulation and process risk assessments from the early stages of the formulation development. The evaluation should encompass formulation components, components of container closure system, and the complete manufacturing process. The early risk assessment provides an anchor that guides the development to mitigate the potential high risks. An example from the manufacturing QRM is shown in Figure 2. The risks shown in Figure 2 are generated automatically based on the values of “Uncertainty (U)” and “Severity (S)” in the early development stage. In late development stage, “Detectability (D)” is also included in the evaluation. The justification for the ranking is provided in the Reference (Ref). Typically, reference is a laboratory investigation report or published literature. Lack of reference (lack of knowledge) will automatically increase the uncertainty and will lead to higher risk. It is mandatory to address the “highs” and mitigate them. It is highly recommended to also address the “mediums,” applying the principle “as low as reasonably practicable” (ALARP). Low risks do not require a follow-up action but the ALARP principle still applies.
Excerpt of QRM evaluation of manufacturing process.
As mentioned in the introduction, international Pharmacopeias accept particulate matter as a known phenomenon. Nevertheless, companies are expected to know their processes well and have them under control. Particle laboratories should therefore provide not only composition of a particulate but also clues about its origin. In the following section, we have highlighted the results that particle laboratories can reliably provide. In addition, application of the analytical techniques is demonstrated.
Particle Life-Cycle Concept
To deal with particle-related topics, Novartis Technical Operations in Unterach established a cross-functional particle life-cycle concept, with a designated cross-functional team consisting of experts from manufacturing, quality, and development. A particle laboratory is not simply an analytical station — it plays a central role in the life cycle, as it also collects and trends particle-related data. This concept involves a comprehensive summary of all process steps in order to understand where particles can enter the system. After defining the process steps, proper actions are defined to reduce particle load. The whole process was divided into five parts: starting materials, manufacturing, visual inspection, AQL testing, and packaging and distribution. In each step, actions were defined to reduce particles; the detailed overview is provided in Table 4. After the processes listed in Table 4 were implemented, the particle life-cycle was established. The term cycle is important, as feedback “to” and “from” the particle laboratory is essential to ensure continuous process improvement.
Particle Life-Cycle Concept at Novartis Technical Operations in Unterach
Phases of an Investigation
Phase 1. Building of Knowledge:
The collection and storing of knowledge is arguably the most important phase. It requires choosing the appropriate analytical methods and focusing on building local expertise. Particle analysis requires — and this differentiates it from “structural elucidation”—handling of miniscule amounts of materials that cannot be isolated in a pure state. Therefore, a comparative approach must be adopted and requires creating a detailed database of spectra from the following materials:
Formulation components. Known or potential degradants/by-products of API and excipients, if available. Contributions from the product development phase result from QRM application.
Materials used during manufacturing and materials present in the manufacturing facility. The first focus is on those materials that have contact with the product (process-related materials). The collection of data is a continuous project to ensure up-to-date information.
Typical materials that can enter into the system or production cycle (e.g., cellulose or other extrinsic particles).
Collecting and building such a library requires cross-functional collaborations of experts of various fields. Libraries encompassing typical chemicals and materials are commercially available and will facilitate the knowledge-building phase, but only an additional internal input and continuous internal enhancement of the data will ensure reliable results. The initial knowledge (basic library, literature data, prior experience, education, etc.) is also further enhanced by all other phases of any particle investigation, and this feedback loop presents the true essence of the QbD approach.
The success of outsourcing of particle analyses to Contract Research Organizations (CROs) depends on the type of particles. Isolated extrinsic and intrinsic particles are identified quite well by qualified CROs, but if the particle is related to the API or excipients, the CROs will often lack the relevant library spectra, and the investigation might stall.
As a long-term investment, we recommend establishing an on-site particle identification laboratory that can support the facility in all particle-related problems. If the expertise is on-site, more reliable results can be provided and in a shorter period of time, compared to external outsourcing of analyses. In case of uncertainties or additional questions, feedback can also be provided quickly.
Phase 2. Identification:
The acquired knowledge allows for identification, which is the main focus of this article. In some cases, one can infer based on solely the morphology of a particle (microscopy) and confirm the results by means of spectroscopic methods if needed. Use of spectroscopic methods has some pitfalls. The results, typically reported as a percentage match between the measured and library spectra, must be critically evaluated considering present and/or absent functional groups, composition of the formulation, and materials used during manufacturing. Sometimes, readjustment of the measuring parameters or refocusing will provide different results. The pertinent question is: Do the results make sense? This expertise differentiates particle analysis from routine measurements, like density or pH measurements. For particles related to formulation components or particles formed by interaction between several components, the analytical result is often not reported unambiguously. For example, “API” or “rubber”, is reported vaguely as “API-related,” “excipient-related,” or “rubber-like” (we refer here to our Raman and IR libraries). In these cases, only a combination of several analytical methods will lead to success. However, even combining various methods may not always provide the desired information, and further considerations are necessary to move on with the investigation. It should be clarified how much information is actually required to solve the original problem, as well as how quickly the information is needed. One of the major limitations of particle investigations is that limited amount of material is available, often only one particle.
One example demonstrated thorough the investigation of particulate matter issues in injectable drug products was presented by Krishna et al. with the drug NeoProfen (21). The interaction of API ibuprofen with glass from the container closure system led to the formation of ibuprofen aluminium-hydroxide particles. These were ultimately identified through a combination of Raman, FT-IR, scanning electron microscopy (SEM)/ energy-dispersive X-ray spectroscopy (EDX), inductively-coupled plasma mass spectrometry (ICP-MS), and analysis of glass surface by X-ray photoelectron spectroscopy.
Phase 3. Find the Source:
The next stage after identification of a particle is to find its source. For extrinsic particles, it is necessary to identify a number of potential sources and evaluate each of them. Could the cellulose that was found originate from cotton, wiping tissues? Is there any paper present in the manufacturing facility? Sometimes both intrinsic and extrinsic sources are possible. Most of the time, the particle laboratory cannot provide unequivocal information about the source, but it can provide clues that direct further investigation. To come closer to the solution, the investigation proceeds through discussion with the appropriate process experts.
The creation of a detailed catalogue of materials present in the manufacturing facility — in other words, process analysis — is recommended. A catalogue entry should preferably contain photo-documentation of a material (overview and microscopy) and analytical results related to its composition. The spectra should be included in respective spectral databases. In the routine work, we include in the database only spectra that differ from previous entries.
An example of data entry into such catalogue is shown in Figure 3. Head covers (hairnet) shown in Figure 3 are an element of the typical gowning requirements worn by production personnel to prevent contamination from hair. Hair protection consists of the following three materials: fabric, thread, and a rubber band. All these components were analyzed separately. The fabric was made of polypropylene, the thread of polyethylene terephthalate, and the band of polybutadiene (rubber). Thus, if a particle is identified as polypropylene, hair protection should be considered as one of the potential sources. Once the source is found, a corrective action will benefit other products as well.
Parts of Catalogue of reference materials of Sterile Production Unterach.
Notes: (a) Hair protection database entry. Fabric: polypropylene (PP), thread: polyethylene terephthalate (PET), band: rubber material. Recorded with camera and Leica M205A stereomicroscope.
(b) Because the rubber material was not clearly identified, an FT-IR spectrum was added to the catalogue. Recorded with Agilent Cary 610-660 spectrometer. PP and PET FT-IR-spectra are not included in the catalogue, but are stored in the spectral database.
In case of intrinsic particles, this stage seeks answers for the following question: What triggered the particle formation? Are there interactions between the formulation components and container closure system or manufacturing equipment that were overlooked? Were there any temperature excursions during transport or storage?
Phase 4. Improve:
Finally, identifying the source of particles allows for improvement for example by:
adjusting the manufacturing process, the formulation, or container closure system;
considering another source of API or excipients or specifying better their key properties; and
re-defining handling or storage conditions of API, excipients, or FDF.
At this stage, close cooperation with regulatory colleagues is required to ensure changes are within the filed scope.
Sample Preparation
At Novartis Technical Operations in Unterach, the initial input for particle investigation is provided by visual inspection of vials in front of a black-and-white board, with defined intensity of light according to Pharm. Eur. (22). The observed particles are described (glass-like, metal-like, fibrous, silicone-like, API-like, etc.), and affected vials are transferred to particle laboratory for detailed investigation.
The sample preparation may be divided into two steps:
opening of a container; and
isolation of the particles.
Care must be taken not to introduce any particles during opening of the container. The work itself is performed in laminar airflow (LAF) workbenches in a “particle-free” laboratory. LAF workbenches are typically working under clean room class grade A (see in Table 3 — maximum permitted number of particles per cubic meter = 3,520 (>0.5 μm) and 20 (>5 μm)). To avoid external contamination prior to opening, the container was cleaned with a particle-free wipe that was moistened with water or aqueous isopropanol. It is known that opening of ampoules is accompanied by the formation and insertion of glass particles, and this fact should be taken into account for particle investigations (23). The surface of vials and stoppers is wiped again after removal of the flip-off and after removal of the crimp-cap. The stopper should not be removed during removal of the crimp-cap to avoid external contamination. All three common techniques of sample preparation are used — extracting a particle by capillary, filtration, and centrifugation (12).
Capillary Technique:
At Novartis Technical Operations in Unterach, the capillary technique is preferred, using which the observed particle is drawn into a micropipette by means of capillary tubes, placed onto a filter, washed, and analyzed. The capillary technique requires only minimal handling, is fast and poses the least risk of external contamination. For the microcapillary technique, the “particle-free” environment of clean room class A is less critical for particles in the visible range (>50 μm). This technique can be well used in less strictly controlled laboratories, for example, in development laboratories at an early development stage.
Filtration:
As per Novartis Technical Operations in Unterach best practice (SOP), if the particles cannot be isolated by a capillary tube, filtration is performed. We routinely use gold-coated filters with porosity of 5 μm. Filtration might require the determination of an initial particle load of the filtration system. This is done by preparing reference blank filters. Purified, particle-free water is filtered, and the resulting blank filters are checked under a microscope or spectrometer. The cleanliness of the system is accepted if the requirements of a specific particle load, described in the SOP, are met. Often, however, a particle can be unambiguously identified (which means it can be recognized by the naked eye, because it has some characteristic morphology, for example, long fiber), blank filters need not be prepared and the liquid can be filtered without checking the particle load.
Following filtration, the emptied vial is rinsed with particle-free water and this liquid portion is also filtered on the same filter. Use of gold- or silver-coated filters is recommended, owing to their weak background signals when using certain spectroscopic methods. The disadvantage of such filters is coloration, which may interfere with microscopic methods. A new filter for each sample is recommended. However, gold- and silver-coated filters can also be regenerated and used repeatedly, which may be acceptable for some development purposes. To regenerate a filter, it is cleaned manually with a particle-free wipe that is moistened with aqueous isopropanol. The cleanliness of the filter is then checked under the microscope or spectrometer; the filter should not contain particles larger than 50 μm, while some scratches on the surface of the filter are acceptable. The recycled filters can also be used if it is clearly traceable which particle was originally observed during visual inspection and the particle can be clearly observed after isolation. Various other filters can be used for the filtration method, for example, 0.2–5 μm polycarbonate filters, 0.22 μm polyvinylidene difluoride (PVDF) filters, 0.22 μm polyethersulfone (PES) filters, 0.22 μm polytetrafluoroethylene (PTFE) filters, or 0.45 μm cellulose nitrate filters are also used at Novartis Technical Operations in Unterach. These filters have a white background and are suitable for determination of optical properties through microscopic methods; however, they are not recommended for spectroscopic use, as the materials they are made of all show vibrational spectra (FT-IR and Raman) and interfere with the analysis.
Filtration is also the method of choice to determine the particle distribution in a vial. In this case, preparation of the blank filters and determination of the initial particle load of the filtration system is mandatory.
Centrifugation:
Centrifugation allows collection of a larger number of particles, provided several containers are pooled. This is necessary for some analytical techniques, for example, NMR or for analysis of isolated redissolved particles by ICP-MS. An approach recently applied by a particulate matter investigating development team included redissolution of centrifuged particles, followed by colorimetric titration to quantitate the iron content (24). In this case, ICP-MS analysis of the vial contents from particle-containing samples suggested a higher content of dissolved iron species.
Light Microscopy
Light microscopy is a classic method to evaluate morphology of particles (14). The microscopic analysis should, however, be accompanied by spectral matching (IR or Raman) or elemental analysis (SEM/EDX or ICP-MS), whenever possible, to reduce uncertainty. Microscopy can direct further investigation — for example, if particles appear shiny and metallic, subsequent analysis by LIBS would be a better option than FT-IR. Analysis of components of the container-closure system or materials that were in contact with the product solution provides important clues for a root cause analysis of particles. Analysis of the glass vial surface should accompany investigations related to glass delamination, while microscopic investigation of damaged stoppers or production gaskets can provide clues about formulation incompatibility. For example, at Novartis Technical Operations in Unterach, particles were observed in one of the products at later development stage. They were identified as silicone particles and attributed to the silicone tubing used in a peristaltic pump. The pump tubing in the peristaltic pump is exposed to high mechanical stress. This mechanical stress and contact of the formulation with the tubing led to rapid abrasion of particles from the tubing (Figure 4). Alternative and more durable pump tubing made of fluoropolymer and silicone was tested in the development, solved the problem and provided feedback into the QbD life cycle. Prior implementation of the change in the production, the new process had to undergo scrutiny discussed in the particle life-cycle chapter and Table 4.
Abrasion on the inner part of silicone pump tubing.
Note: Recorded using Leica M205A stereomicroscope.
In case of analysis of single particles, light microscopy is used routinely to determine basic optical properties (e.g., color, shape, and morphology) as well as more sophisticated optical properties (e.g., refractive index, birefringence, and interference contrast). In some investigations, light microscopy provides enough unambiguous data, a typical example for that is distinguishing clear and amber glass particles.
A long-term aim of pharmaceutical companies is to obtain sufficient information from light microscopy methods to identify the particulate composition so that no further analysis is required (saving time and resources). To carry out identification based on microscopic data, the process analysis results should be accompanied by representative photo documentation to be later capable of identification by matching.
Raman Spectroscopy
At Novartis Technical Operations in Unterach, the main technique for analysis of particulate matter is Raman spectroscopy (25). Raman spectra are unique for individual compounds and allow quick matching with references from the spectra library. Using a Single Particle Explorer, no physical contact occurs between the light source and the particle, so the chance of losing or damaging the particle during the analysis is minimized.
The Single Particle Explorer also possesses a counting function: it is capable of an automatic measurement to determine the number of particles on a filter and their size. After the automatic measurement, we manually focus on individual particles and identify them using database comparisons (Figure 5).
Raman spectra of a silicone particle.
Notes: Red: raw spectrum; green: reference spectrum of best database match; blue: background-corrected spectrum. Ranking: 989 (1,000 would mean 100% match). Recorded with Single Particle Explorer; S/N means signal-to-noise ratio.
An important feature of Raman spectroscopy is that the method is insensitive to the presence of water (in contrast to IR). This is particularly useful for aqueous liquid injectable formulations. The practical advantage here is that the isolated particles do not have to be dried, and thus, potential external contamination is limited. Cao et al. even reported in situ particle identification (i.e., they identified particles directly in glass containers without isolation) (26). Not all compounds are good scatterers (glass, amorphous materials) and in case of dark materials, fluorescence is a problem, but it can be somewhat reduced by using a preburning function (energy transfer through the laser to reduce fluorescence in transmission Raman spectroscopy).
Fourier Transform Infrared Spectroscopy
Infrared spectroscopy has been used for both qualitative and quantitative analysis for decades (15, 27). Its particular strength lies in identifying functional groups of molecules. Particle analysis leverages uniqueness of the IR spectra by matching the sample with the library spectra. Attenuated total reflection (ATR) is the preferred Fourier transform Infrared (FT-IR) technique nowadays. Performing ATR measurement does not require any sample preparation, is facile and has essentially replaced the traditional transmission methods (cell, NaCl, KBr), also including the analysis of visible particulate matter. The downside of the ATR is that the sample must be in direct contact with the ATR-crystal that might damage the particle. ATR is not a trace technique and requires a sample surface of at least 10 μm2. The refractive index of the crystal must be high, typically 2.38–4.01. Diamond, germanium, or zinc selenide crystals are mostly used as ATR crystals. Single particles can be analyzed using ATR FT-IR microscope. In some cases, such as massive precipitation, simple ATR units without microscope can also be used. Library or peak searches coupled with modern software allows analysis of particle mixtures. Detailed discussion of potential errors encountered during interpretation of IR spectra was provided by Chalmers (28). An example using FT-IR as the main tool for investigation of particles in a formulation is discussed in a case study on the diuretic drug conivaptan hydrochloride by Ban et al. (29). An increase of pH over time led to precipitation of the corresponding free base analyzed by FT-IR and SEM/EDX.
A combination of orthogonal techniques such as Raman and FT-IR methods is particularly useful. The two complement each other and the analysis can be done successively on the same filtered particle. A compound that is Raman-active is often IR-inactive and vice versa (Table 5). Thus, the sample prepared on a gold filter and analyzed by Raman can be subsequently subjected to FT-IR analysis. Provided both instruments are ready for use, such sequence takes merely few minutes. An example for characterizations, in which FT-IR spectra were superior to Raman spectra, was includes in the analysis of particles attributed to rubber stoppers used in Novartis Unterach. The material of these stoppers consists mostly of poly(chlorobutyl)rubber that does not provide Raman spectra of sufficient quality. However, FT-IR provided spectra with good signal-to-noise ratio and allowed identification of the particles using the database comparison (Figure 6).
Comparison of Raman and IR Methods
FT-IR spectrum and database comparison of a rubber stopper particle.
Notes: Left: spectrum of a particle; Right: overlay of the particle spectrum (black) with the reference spectrum (red). Recorded with Agilent Cary 610-660 spectrometer.
Laser-Induced Breakdown Spectroscopy
LIBS is a spectroscopic technique method, which is also useful for investigations related to the analysis of particulate matter. In this method, a laser is focused on the test sample, resulting in plasma formation that emits characteristic atomic emission lines specific to their constituent elements. LIBS is, therefore, a type of atomic emission spectroscopy and the technique is usually applied for identification of metal particles or trace elements. It provides elemental analysis–related information. Organic compounds consist of C, O, or sometimes N with various bonding possibilities, so no unequivocal conclusion related to structure can be drawn from LIBS. Capabilities and limitations of the technique were discussed by Cremers et al. (30, 31). LIBS provides equivalent data to elemental analysis. However, having a Single Particle Explorer with a spectral database, a comparative approach (to the database) is applied, because it is faster and more precise.
The disadvantage of LIBS is that it is a destructive method, and most of the time the particle will be destroyed. Because only single particulates are often analyzed, the application of LIBS has to be carefully considered. A typical application of LIBS is for particles that appear metallic under the microscope. An example showing the LIBS spectrum of a particle identified as steel is shown in Figure 7.
LIBS spectrum of steel (“Stahl”) particle.
Notes: Red: raw spectrum; green: reference spectrum of best database match; Blue: background-corrected spectrum. Ranking: 884 (1,000 would mean 100% match). Recorded with Single Particle Explorer; S/N means signal-to-noise ratio.
Supporting Methods
Supporting methods are typically not used alone but in combination with other techniques. This categorization is, however, purely subjective.
Scanning Electron Microscopy Coupled With Energy-Dispersive X-ray Spectroscopy
SEM coupled with EDX provides both morphological and chemical composition information for a particle (32). The sample preparation is minimal. Filtered samples can often be directly analyzed without further manipulation. For particle analysis, EDX is used mainly for qualitative analysis to identify elements present in the samples. Depending on the instrument setup, elements with molecular weights from beryllium up to uranium are detected. Quantitative analysis is often not possible due to sample inhomogeneity.
Most of the particles in the pharmaceutical industry are organic materials. For such non-conductive materials, coating with gold or carbon is required to improve the refractive properties of the particle. For SEM-EDX, a comparative spectral database is helpful. However, organic matter cannot always be unequivocally identified based on spectral matching. This is due to potential differences in the measuring parameters, equipment (sample holder), or due to sample inhomogeneity. A good hint to particle characterization is determination of the trace elements of the possible reference materials, such as filters used and production-supporting materials.
Inductively-Coupled Plasma Mass Spectrometry
ICP-MS has been primarily used to analyze elemental composition of liquid samples (33). As the method allows analysis of multiple elements, it has typically been used in extractable and leachable studies (34). With the introduction of the new Q3D guideline for elemental impurities, the method is becoming more routine. For particle analysis, there are two applications:
elemental analysis of isolated and redissolved particles to understand their composition;
analysis of the parent formulation itself to screen for an elevated number of elements that might have contributed to the formation of particles.
Other Techniques
Other techniques can be used on a case-to-case basis for the purpose of investigation. Mass spectroscopy (MS) and NMR spectroscopy have been used by Novartis particle-investigating teams, in search of API-related particles. Shearer recommended the use of transmission electron microscopy (TEM) as an alternative to SEM to analyses thin glass flakes in investigations related to glass delamination (35).
Particle Mapping
As particles are an undesired phenomenon, their presence essentially questions the following:
robustness of the manufacturing process;
formulation understanding.
Checking robustness of the process is relatively easier of the two, as it essentially requires tracking and evaluating process changes. For example, reviewing cleaning procedures, sterilization processes, exchanges of filling needles, tubing and gaskets. Investigations should typically find the root cause of the particle formation or exclude process as the root cause.
Gaps in formulation understanding are a challenging area. They include reactivity and/or compatibility issues that were not considered, were unknown or were not observed during development and early stability studies. Typically, it is very difficult to identify the structure of API-related particles such as degradation products or complexes. The fact that most of the particle-related investigations remain unpublished contributes to relatively slow progress in understanding particle formation. A simplified map of the most common root causes to consider is shown in Figure 8.
Potential root sources of particle formation.
What makes prediction difficult and root causes for particle formation surprising is that in reality, the web is much more interconnected than shown in Figure 8. An example being the relationship between pH and leachables. A chain of events such as the following could lead to a particle event:
Chelator → Container closure system → Leachable → Change of pH → Solubility of API → PARTICLE
Elemental impurity → Reactivity of API → Incompatibility with excipient → PARTICLE
A priori it is difficult to predict such particle formation even in the QbD approach. Reactivity understanding of API and excipients, including their complexation behavior, is one of the key components to be considered. Internal extractable and leachable studies from the container closure system as well as supplier information are extremely helpful. For example, Borchert's extractable study from borosilicate glass containers showed different leaching propensities of certain elements at various pH. Aluminum and calcium essentially did not leach at pH 4–8, but leaching increased dramatically at approximately pH 9.5 (36). The presence of a chelator in the formulation might further exacerbate leaching of certain elements. General practical fundamentals of glass, rubber, and plastic sterile packaging systems were summarized by Sacha et al. (37).
Example and Critical Evaluation of Results
An example demonstrating how settings of an instrument can influence results and importance of critical evaluation of the results is discussed in the following case study. During cleaning before routine production, a yellowish particle was found in a vessel (see Figure 9). The particle was collected and transferred to the particle laboratory with a characterization request. It was placed on a gold-coated filter, analyzed by Raman spectroscopy, and identified as “white petroleum jelly” on the basis of a database search. This result indicated the presence of a polymer with long-chain hydrocarbon structure (see Figure 10, left spectrum), which could mean a degraded or modified organic compound, deposited on the surface of the particle. The measurement parameters were re-adjusted, repeated at the same spot so that the Raman laser penetrated deeper into the particle (i.e., below the modified surface). The particle was then identified as poly(ethylene terephthalate) (PET) with very high (>95% match) ranking (Figure 10, right spectrum). This suggested the presence of a PET-particle, which was degraded during the process, and therefore, long-chain hydrocarbon structure compounds could be observed on the surface. PET could have originated from cleanroom towels and cleanroom clothing.
A yellow particle found in a vessel before production during cleaning.
FT-IR spectra of the original particle (left spectrum) and the particle below the surface near the core structure (right spectrum).
Conclusion
In order to control and reduce the occurrence of particulate matter, Novartis Technical Operations in Unterach established a particle laboratory. The laboratory analyzes materials of unknown composition and provides support to process experts and investigators. The information obtained from studies in the particle laboratory helps to identify the origin of the detected particulate matter and support root cause analyses. The particle laboratory can, after characterization, categorize whether the particulate matter is inherent, intrinsic, or extrinsic. The isolation techniques and analytical methods are listed, in order of their preference, in Table 6. The adopted particle program at Novartis led to improved process understanding and resulted in a decrease of reject rates, AQL failures, and corresponding batch rejections. The ultimate goal has been to provide the patients with medicines of uncompromising quality.
Overview of the Particle Analysis
Conflict of Interest Declaration
The authors declare that they have no competing interest.
Acknowledgments
The authors thank Robin Daly, Robert Eskes, Martine Heeren, Kai Kramer, and Christian Pflügl for helpful discussions and support during preparation of the manuscript.
- © PDA, Inc. 2018
