Application of ICH Q9 Quality Risk Management Tools for Advanced ...

25 downloads 89590 Views 2MB Size Report
Nov 11, 2016 - risk management approach optimizes the product development sustainability ... Administration encourages the application of Quality by Design.
Journal of Pharmaceutical Sciences 106 (2017) 278-290

Contents lists available at ScienceDirect

Journal of Pharmaceutical Sciences journal homepage:

Pharmaceutics, Drug Delivery and Pharmaceutical Technology

Application of ICH Q9 Quality Risk Management Tools for Advanced Development of Hot Melt Coated Multiparticulate Systems Elena Stocker 1, Karin Becker 1, Siddhi Hate 2, Roland Hohl 3, Wolfgang Schiemenz 4, Stephan Sacher 3, Andreas Zimmer 1, Sharareh Salar-Behzadi 3, * €tsplatz 1, Graz 8010, Austria Department of Pharmaceutical Technology, Institute of Pharmaceutical Sciences, University of Graz, Universita Department of Industrial and Physical Pharmacy, Purdue University, Heine (Robert E.) Pharmacy Bldg, 575 W Stadium Ave, West Lafayette, Indiana 47907 3 Research Center Pharmaceutical Engineering (RCPE) GmbH, Inffeldgasse 13, Graz 8010, Austria 4 Hermes Arzneimittel GmbH, Großhesselohe, Germany 1 2

a r t i c l e i n f o

a b s t r a c t

Article history: Received 9 April 2016 Revised 20 August 2016 Accepted 26 September 2016 Available online 11 November 2016

This study aimed to apply quality risk management based on the The International Conference on Harmonisation guideline Q9 for the early development stage of hot melt coated multiparticulate systems for oral administration. N-acetylcysteine crystals were coated with a formulation composing tripalmitin and polysorbate 65. The critical quality attributes (CQAs) were initially prioritized using failure mode and effects analysis. The CQAs of the coated material were defined as particle size, taste-masking efficiency, and immediate release profile. The hot melt coated process was characterized via a flowchart, based on the identified potential critical process parameters (CPPs) and their impact on the CQAs. These CPPs were prioritized using a process failure mode, effects, and criticality analysis and their critical impact on the CQAs was experimentally confirmed using a statistical design of experiments. Spray rate, atomization air pressure, and air flow rate were identified as CPPs. Coating amount and content of polysorbate 65 in the coating formulation were identified as critical material attributes. A hazard and critical control points analysis was applied to define control strategies at the critical process points. A fault tree analysis evaluated causes for potential process failures. We successfully demonstrated that a standardized quality risk management approach optimizes the product development sustainability and supports the regulatory aspects. © 2016 American Pharmacists Association®. Published by Elsevier Inc. All rights reserved.

Keywords: lipids infrared spectroscopy formulation fluid bed factorial design coating dissolution oral drug delivery particle size

Introduction In 2004, the U.S. Food and Drug Administration published its final report on pharmaceutical quality and guidance for the industry on process analytical technologies, introducing a new regulatory framework for development of pharmaceutical manufacturing with enhanced product quality.1-3 The ambition was to increase process capability, product development, and manufacturing efficiency, and to reduce product variability by improving product and process design, scientific understanding, and control strategies.4 For this reason, the U.S. Food and Drug Administration encourages the application of Quality by Design (QbD) as a systematic holistic scientific and risk-based approach in drug product development, manufacturing, and regulation by

Elena Stocker and Karin Becker contributed equally as first author. * Correspondence to: Sharareh Salar-Behzadi (Telephone: þ43-316-873-30948; Fax: þ43-316-873-1030948). E-mail address: [email protected] (S. Salar-Behzadi).

identifying critical product characteristics and applying tools such as quality risk management (QRM).4-7 With the International Conference on Harmonisation Q9 guidance and the International Organization for Standardization (ISO) standards such as ISO 14971, ISO 31000, International Electrotechnical Commission 31010, ISO 73, and IS/TR 31004, a selective and prospective QRM was provided for pharmaceutical research and development, production, and clinical testing.8-13 Furthermore, the ISO/International Electrotechnical Commission Guide 51 defines risk as a combination of severity of harm and the associated probability of its occurrence and provides a guideline to reduce risk to a tolerable level.14 The design and development of a robust drug product requires serious consideration of the physical (e.g., particle size distribution, polymorphism, and melting point), chemical (e.g., pKa, solubility, and stability), and biological (e.g., partition coefficient, bioavailability, and permeability) characteristics of a drug substance and of the selected excipients as well as their potential interactions.4 These characteristics of the input materials (drug, excipients, intermediates) are defined as material attributes (MAs) and have a potential impact on the quality of the intermediate or finished drug 0022-3549/© 2016 American Pharmacists Association®. Published by Elsevier Inc. All rights reserved.

E. Stocker et al. / Journal of Pharmaceutical Sciences 106 (2017) 278-290


Figure 1. Step-by-step description of the QRM workflow including risk assessment, control, and review of the development of HMC process.

product.4,15 The quality target product profile (QTPP) summarizes all quality attributes (QAs) of the output materials or drug products that should be achieved to ensure the desired quality by taking into account safety and efficacy of the drug product.4,16 Further information included in the QTPP is the route of administration, the dosage form and strength, the container closure system, the pharmacokinetic characteristics, and the stability of the drug product.4 Critical quality attributes (CQAs) have been defined as the physical, chemical, biological, or microbiological properties or characteristics that should be within an appropriate limit, range, or distribution to ensure the desired product quality.4,7 Risk assessment and scientific knowledge help to prioritize the critical material attributes (CMAs) posing a high risk to product quality, to identify the CQAs and the critical process parameters (CPPs) and to link CMAs and CPPs to CQAs.4 Viable tools for this linkage and the identification of CPPs in pharmaceutical unit operations are the standard QRM tools, for example, the failure mode effects analysis (FMEA), cause and effect analysis (Ishikawa fish bone diagram), fault tree analysis (FTA), or hazard analysis and critical control points (HACCP).13,17 Control charts, Pareto charts, or design of experiments (DoE) are mentioned in The International Conference on Harmonisation Q9 guideline as supporting statistical tools for a systematical operating approach within a predefined processing space based on prior knowledge.13 Several pharmaceutical studies have been published using QbD and QRM to provide a risk-based and proactive approach for the

development of robust formulations and the optimization of pharmaceutical processing.18-22 Most of these studies applied FMEA, Ishikawa fish bone diagrams, or other tools such as qualitative initial risk-based matrix analysis in combination with statistical design to identify CPPs and CQAs and optimize various pharmaceutical unit operations.18-22 Moreover, a biopharmaceutics risk assessment roadmap has been introduced as a powerful patient-centric approach for optimizing product development and performance. In this approach, therapy-driven target drug delivery profiles are used as a framework to achieve the desired therapeutic outcome, by considering the clinical relevance in the early formulation development stage and using biopharmaceutical tools to identify the potential challenges of product optimization for patient benefit.23 In the last years, applying in silico techniques has gained major attraction in the industry and academia for estimation of pharmacokinetics, pharmacodynamics, and toxicity parameters of compounds and formulations of interest.23,24 Moreover, several theoretical mechanistic process modeling techniques such as computational fluid dynamics, direct numerical simulations, or the discrete element method have been developed to significantly reduce the number of required experiments.25 For example, Stocker et al. and Adam et al. successfully integrated a discrete element method simulation within the framework of QbD and QRM and improved the scientific process understanding of a tablet coating and a blending unit


E. Stocker et al. / Journal of Pharmaceutical Sciences 106 (2017) 278-290

operation significantly.26,27 After a detailed risk assessment, it is recommended to control risks by implementing an action plan, which intends to improve the detectability of harm or to mitigate the severity and probability of hazards for the drug product quality to an acceptable level.13 Such actions can involve facility, engineering, or process design changes, training or the enhancement of product performance monitoring through the implementation of process analytical technology tools.17 Regular risk monitoring and reviewing ensure continuous re-evaluation of risk control activities, risk levels, and existing control measures.28 An interesting example for the application of QRM is the hot melt coating (HMC) process. This technology has aroused a recent interest in the pharmaceutical industry for producing microcapsules.29-31 HMC is a solvent-free coating technology, in which molten lipid-based excipients are applied as coating materials.32 HMC offers several advantages over solvent-based coating processes such as the omission of time-consuming and costly solvent evaporation and recovery steps and the use of cheaper and safer green excipients.32-35 This work is the first published approach for applying QRM to control an HMC process as a pharmaceutical manufacturing technique. The intermediate product of interest was a hot melt coated multiparticulate system. The coating was used for masking the unpleasant taste of the active pharmaceutical ingredient (API) and providing a stable immediate release profile. A target quality specification for the intermediate product and the CQAs of interest were defined and applied for the QRM of the manufacturing process. The workflow presented in Figure 1 shows the applied QRM tools. Materials and Methods Material N-acetylcysteine (N-ac), a mucolytic agent with the daily dose of 600 mg, was purchased from PharmaZell (Raubling, Germany) and used as core material. The coating consisted of tripalmitin and polysorbate 65. Tripalmitin was kindly provided by IOI Oleo (Hamburg, Germany) and polysorbate 65 was obtained from Croda (Nettetal, Germany). All chemicals were of analytical grade and purchased from Sigma-Aldrich (Darmstadt, Germany). The nylon membrane filters and the mixed cellulose ester membranes were obtained from Merck (Darmstadt, Germany) and Merz Brothers (Ansfelden, Austria), respectively.

Hot Melt Coating Process The HMC process was performed in an Innojet VENTILUS® V-1 laboratory system with an IHD-1 hot melt device (Romaco Innojet, Steinen, Germany). The temperature of the melt and the atomization air were set to 100 C. The inlet temperature was set to 25 C, causing a product temperature of 34 ± 2 C preventing the degradation of N-ac and reducing the agglomeration risk during coating. The product temperature was continuously recorded with an infrared sensor. A batch size of 300 g N-ac was used for all experiments to ensure homogenous fluidization during processing. The fluidization of the core material was stabilized before starting the coating process. During coating, the molten coating material was sprayed on the fluidized particles by using defined spray rates and atomization air pressures. After screening the applicable range of process parameter settings, a statistical DoE was created using the software MODDE 10 €, Sweden) to investigate the criticality of the (Umetrics, Malmo defined process parameters and MAs on the CQAs. The defined CQAs were (1) taste masking efficiency, (2) immediate release profile, and (3) particle size. Taste masking was specified as efficient, when 0.4% of N-ac was released from the coated particles after 1 min of dissolution testing. The defined specification limit for an immediate release profile was 85% wt/wt API release within 30 min of dissolution testing. The particle size of the coated product was defined as acceptable if D90 was 820 mm. Because a finished product will contain a matrix of sweeteners and flavors additional to the coated API particles, a widening of the specified acceptance criteria for taste masking is justifiable. The matrix supports taste masking and therefore a maximum release of 20% of N-ac after 1 min of dissolution testing may be appropriate to ensure proper taste masking. Because the finished product is not the focus of interest in this article, this specification parameter will not be discussed.

Determination of Release Profile of API From Microcapsules Dissolution tests were performed in triplicates with a dissolution tester DT 820 LH United States Pharmacopoeia-2 (Erweka, Germany) applying a paddle stirring speed of 100 rpm. Volumes of 900 mL of 0.1 N hydrochloric acid [pH 1.0 (±0.1), temperature 37 C (±0.5 C)] were used as dissolution medium. Coated samples containing approximately 600 mg of coated N-ac were subjected to dissolution testing. One milliliter of the dissolution medium was automatically sampled after 1, 5, 10, 15, 20, 30, 45, and 60 min and

Table 1 Procedure Steps and Description Covered in the HACCP Plan HACCP Procedure Step


Description of Theoretical Case (result of this study)

Process step Quality hazard

Critical process step Problem or situation that has an impact on the product quality or the process Measures and control systems to detect the hazard or to reduce the probability of occurrence Where in the process is the control measure executed? What is measured? Which result would be ideal? The allowed range of the measurement results to ensure the product meets the required specifications With which method is the critical control point monitored and how are the results documented? Actions to be carried out to keep results within the accepted range/to return to the accepted range if the critical limits are not met Methods to verify the effectiveness of the control measures and the corrective actions Documentation of decisions and corrective actions used

Coating Coating amount

Preventive measures Control point Critical limits Monitoring procedure Corrective action Verification Record

Off-line HPLC analysis, in-line NIR spectrometry Coating amount [% (wt/wt)], particle size enlargement Acceptance criteria: 25-40 [% (wt/wt)], coefficient of variation, particle size (D90  820 mm) Off-line HPLC analysis to determine the amount of API, quality control of the final drug product Adaptation of spray rate, atomization air pressure, air flow rate within defined process ranges Off-line and in-line NIR measurements: amount of coating on particle Manufacturing record and decision tree for HMC

E. Stocker et al. / Journal of Pharmaceutical Sciences 106 (2017) 278-290

filtered using a 0.22 mm membrane filter. The withdrawn volume was not replaced, but considered in mathematical measurements.

Content Assay The exact N-ac content after coating was determined by grinding 6 g of the coated particles in a CryoMill (Retsch, Germany) for 5 min, with an agitation frequency of 25 1/s and nitrogen cooling to 196 C. An amount of the grinded powder comprising 600 mg N-ac was precisely weighed into a volumetric flask, dispersed in 100 mL of phosphate buffer pH 6.8 (±0.1) and treated in an ultrasonic bath for 15 min in order to destroy coating and to dissolve the complete amount of N-ac. Dispersion was filtered through a nylon membrane and analyzed by HPLC.

HPLC Measurements The HPLC measurements were performed using a Waters 2996 PDA Detector 195 HPLC system with an autosampler and a Synergi Fusion RP 4 mm column (80 Å, 250  4.6 mm) with an Atlantis® T3 (5 mm) pre-column. The mobile phase was a solution of 5% vol/vol acetonitrile in water, and the pH was adjusted to 1.6 (±0.1) with phosphoric acid. The injection volume was 20 mL, the flow rate was 1 mL/min, the column temperature was 21 C, and the autosampler temperature was 5 C. Each sample was analyzed, with a run time of 20 min and detected at a wavelength of 220 nm.

Table 2 List of QTPP of Coated N-ac Particles (intermediate product), CQAs, and Selected CPPs and CMAs for Detailed Investigation Using a Statistical DoE Variable QTPP element Mode of administration Intended final dosage form Dosage strength Pharmacokinetics Container closure system Stability CQAs Dissolution ratea Taste maskinga Particle sizea

API identity Assay Impurities

Target Value Oral administration Multiparticulate system 600 mg N-acetylcysteine Immediate release Sachet (laminated aluminum paper foil) 36 months shelf life (25 C, 60% RH) Immediate release: N-ac release after 30 min 85% N-ac release after 1 min 0.4% 820 mm: NMT 10%a Positive 95%-105% of defined target value N,N0 -Diacetyl-L-cystine: NMT 0.5% N,S0 -Diacetyl-L-cysteine: NMT 0.5% L-Cysteine: NMT 0.5% L-Cystine: NMT 0.5% Not identified impurities: NMT 0.2% Total impurities: NMT 1.0%

CPPs selected for detailed investigation using a DoE Process

Process Parameter

Hot melt coating

Air flow rate Atomization air pressure Spray rate

CMAs selected for detailed investigation using a DoE Process

Material Attribute

Hot melt coating

Coating amount Emulsifier content

NLT, not less than; NMT, not more than; RH, relative humidity. a CQAs discussed in more detail within this work.


Offline Particle Size Analysis A high-speed analysis sensor (QICPIC; Sympatec GmbH, Clausthal-Zellerfeld, Germany) with a RODOS/L dry disperser was used for the particle size distribution analysis. The measured particle size range was 20-3000 mm. The feeding rate was set to 30%, 400 frames per second were taken, the injector diameter was 4 mm, and the air pressure was 1 bar. D90 was determined for the N-ac raw material and for the samples after HMC. The gain in particle size of N-ac crystals after coating was considered as coating thickness (mm). Quality Risk Management Product Design-Failure Mode and Effects Analysis Failure mode and effects analysis (FMEA) methodology was applied to detect potential failures, which most likely occur in the early product design, by considering the potential CQAs.36-38 The analysis focused on deficiencies, which might lead to critically insufficient drug efficacy or lack of patient compliance. Product design (PD)-FMEA uses a risk priority number (RPN) system for a quantitative or qualitative classification of the severity (S), occurrence (O), and detectability (D) of potential failure modes. Furthermore, it is utilized to define and prioritize corrective actions. The risk assessment catalog for a quantitative classification based on a 3-level score (1, 5, and 9) was used in this study for each risk factor,37 ascending the score with increasing criticality. An average RPN value of 125 was defined as the critical limit.26 Please note that there are different ways to define the values of S, O, and D. The guidelines are generally divided into qualitative and quantitative numerical scales. While a qualitative guideline follows expected (theoretical) behavior of the component, a quantitative guideline is specific and follows available data for evaluation.37 Companies may use different guidelines based on the requirements and defined product specifications. Cause and Effect Analysis All CPPs of the HMC process having a potential impact on the CQAs were identified based on a detailed and systematic cause and effect analysis. The results were visualized using an Ishikawa39 diagram, also known as fishbone diagram. Process Flowchart: Characterization of the HMC Process The HMC process was characterized via a flowchart, based on the identified CPPs, with critical impact on the defined CQAs. This flowchart was provided based on the EN ISO 10628.40 Process Failure Mode, Effects, and Criticality Analysis Within a risk-based process analysis, the failure modes that were linked to failure causes and effects of potentially CPPs, MAs, and equipment design parameters were evaluated.41 The CPPs identified in the cause and effect analysis were classified according to their criticality and risk priority, and served for the development of a DoE and a control strategy. The risk factors severity (S), occurrence (O), and detectability (D) were ranked based on a defined assessment catalog using a 5-level scale of odd numbers from 1 to 9 for each risk factor.10 The process failure mode, effects, and criticality analysis (PFMECA) defines an RPN for each potential failure mode of a specific process parameter (PP), which is used for the quantitative evaluation of the process risk priority level and required corrective actions. The combination of probability of failure modes and severity of their consequences are classified by the analysis of criticality. In this work, PPs were classified as CPPs, well-controlled critical process parameters (WC-CPPs), and general GPPs. These 3


E. Stocker et al. / Journal of Pharmaceutical Sciences 106 (2017) 278-290

categories were adapted according to the “a-mab case study.”42 Both defined CPP and WC-CPP were screened using a statistical DoE. A CPP has an impact on QAs and consequently a direct impact on product quality. It must be controlled tightly, because of its limited robustness. A WC-CPP has an impact on QAs, but it is well controlled and can be characterized by robust monitoring. A general PP is an adjustable parameter that has no significant effect on product quality or process performance.

Statistical Design of Experiments The impact of CPPs and CMAs on the defined CQAs was screened using a statistical DoE, created with the software MODDE 10.0 (Umetrics). A fractional factorial design with a resolution of 5 was performed. This DoE consisted of 19 runs containing 3 replicates as center points to evaluate the main and 2-factor interactions of the factors. The main factors were spray rate (2-8 g/min), atomization air pressure (0.8-1.4 bar), air flow rate (30-45 m3/h), coating amount (25%-40%), and emulsifier content (polysorbate 65, 10%-20%). The batch size was 300 g of N-ac for all trials. The inlet air temperature and the atomization air temperature were kept constant at 25 C and 100 C, respectively. The responses were defined as coating thickness (290 mm, concerning the X90 ¼ 532.20 ± 0.8 mm of N-ac crystals without coating and the specification of coated particle size of X90  820 mm), amount of API released (% wt/wt) within 1 min of dissolution test (indicator for taste masking efficiency), and the required time to release 85% (% wt/wt) API (indicator for immediate release profile). The obtained results were used for the evaluation of RPN and reassessment of the PFMECA.

Hazard Analysis and Critical Control Points HACCP is characterized by its hazard control utility at critical points in the pharmaceutical manufacturing process. A critical

control point (CCP) is defined as the step at which control must be applied.43,44 In this study, HACCP was used for the development of preventive process monitoring solutions via a decision tree (Table 1).45,46 Implementation of a process control strategy was evaluated. Furthermore, a concept for precautions due to deviations from specified CCP limits was established.46 Fault Tree Analysis FTA generates process diagrams that help determine the structure of problems and their importance.47,48 The starting point is the top fault event, from which the fault tree works backwards, considering all possible causes (e.g., operator, control system, equipment, maintenance, and environment failure). In this study, the fault tree was used as a tool for root cause analysis based on risk assessment. This supports a detailed process characterization by focusing on causal factors and deviations within the overall control strategy.49 The qualitative FTA generated in this study does not include the calculation of fault rate and probability. Results and Discussion The defined QTPP of the intermediate product is listed in Table 2. The CQAs of the intermediate product with expected significant impact on the QTPP, that is, the particle size and the dissolution profile accounting for immediate release and taste masking efficiency, are also included in this table and evaluated below. Table 2 lists also the CPPs and CMAs studied in more detail in this study. Product Design-Failure Mode and Effects Analysis Table 3 summarizes an initial assessment of the potential failure modes, considering their impact on the defined CQAs. The potential failure modes are defined as follows:

Table 3 PD-FMEA for Potential Failure Modes (A), (B), and (C): Initial Assessment of Severity, Probability of Occurrence, and Detectability of Potential Failure Effects and Impact on CQAs Potential Failure Mode

Potential Failure Effect

Severity Potential Cause(s) for D90 > (S) 820 mm

Occurrence Current Control to Detect (O) Failure

Detectability RPN (D)

(A) Size of hot melt coated particles is too large and expected to stimulate mastication (specification limit: D90  820 mm)

Impact on potential CQAs





5 5 5 9 9

225 225 225 81 81

Potential Failure Mode

Potential Failure Effect

(B) Taste masking is insufficient Impact on potential CQAs and causes unpleasant sour taste during salivation (specification limit: N-ac release after 1 min 0.4%)

Atomization air pressure too low Spray rate too high Product temperature too high Coating amount too high Air flow rate too low Emulsifier content too high

5 5 5 1 1

Off-line particle size measurement with QICPIC (D90)


Potential Cause(s) for API Release >0.4% after 1 min


Current Control to Detect Failure




Atomization air pressure too high/low Emulsifier content too high Spray rate too high or too low Coating amount too low Air flow rate too low or too high Product temperature too low or too high


Off-line dissolution test with USP apparatus II with focus on drug release after 1 min



9 5 5 9

405 225 225 81



5 5 5 1 1

Potential Failure Mode

Potential Failure Effect


Potential Cause(s) for API Release