The use of Real-Time At-Line Particle SizeMeasurement on a Pelletization Process for AccurateEnd-Point PredictionAuthors: Emmet HoganA, Mike MulcahyA, Dr. Ian JonesB, Dr. Norbert PöllingerC, Mirko NowakC, Dr OrapinRubinoD, Greg JayneDAInnopharma Technology, Dublin, IrelandBInnopharma College of Applied Sciences, Dublin, IrelandCGlatt GmbH, Binzen, GermanyDGlatt Air Techniques, Ramsey, NJ, USAAbstractThe direct pelletization process is a fast and flexible operation with the potential to produce high quality particlesappropriate for sophisticated oral solid dose formulations where a narrow Particle Size Distribution (PSD) isrequired. Particle size is a critical quality attribute for many pharmaceutical applications making control of endpoint particle size a fundamental requirement. The Glatt proprietary Complex Perfect Spheres (CPS )pelletization process is fast and therefore requires a rapid objective PSD measurement method duringprocessing to follow the particle growth curve and accurately determine end-point. Traditional methods are tooslow or subjective and therefore may be an unacceptable risk for use as an In-Process Control (IPC) in a controlstrategy. An at-line direct imaging method with automated high-speed image analysis can be practically usedduring processing to accurately predict the process end-point with fast objective PSD data, supported byparticle images, to significantly improve process optimisation and control. The approach can also be used forother rotor based pelletization processes.IntroductionDirect pelletizing allows for a variety of drug formulations to be realized in extremely low to very high doses. Thetechnology is particularly suited to oral dosage forms where gastric resistant and modified release is a priority.Furthermore, a relatively narrow PSD is important to achieve a consistent predictable drug release profile incoated pellets. The CPS technology allows for themanufacture of functionalizedpellets/micropellets with perfectroundness and smooth surface.The CPS process is fast rediction of end-point based onparticle diameter and sizedistributioniscritical.Traditionally, a microscope-basedimage capture and analysissystem is used to measure inprocess samples to predict theend-point. An area within thesample tray is selected and a smallnumber of particles are manuallyImage 1:Infographic of Pellitization with real time IPC 2018, Innopharma Technology 1 of 10

measured and analysed by the operator using the system’s software. Limitations to the use of this in-processtest include a reliance on an experienced process operator to select which particles are analysed combined withthe small number of particles analysed per process sample. This approach might be assessed as relatively highrisk as a process control strategy and could be mitigated by a more objective and rapid in-process test methodto more reliably predict process end-point.The Eyecon2 is a non-product contact Process Analytical Technology (PAT) tool used for monitoring particle sizein-line or at-line, in real-time. The Eyecon2 is a direct imaging camera system which provides PSD data and colourimages, allowing for a deeper understanding of the key & critical process parameters for a given process/system.The data can be used in conjunction with process parameter optimization and quality control (QC) results todevelop a more efficient, robust and reliable process, and deliver a more consistent improved level of processcontrol and therefore product quality.Objectives & PreparationThe objective of this trial is to monitor and record particle growth during a direct pelletization process run. Theprocess runs over a relatively short time period and particle size is a Critical Quality Attribute (CQA) of theproduct. The PSD measurement method should: Allow systematic and consistent sample preparation. Provide short analysis time. Be easy to operate. Provide PSD data and images of the particles. Track the change in particle size during processing. Provide fast and accurate PSD data to predict the process end-point. Demonstrate that the pellets produced are within the desired PSD range. Demonstrate applicability for process development and for process control.The Eyecon2 is usually configured as an in-line particle analyser, interfaced directly on process equipment tomeasure particle size through a viewing port. There is potential for such a configuration on the pelletizationprocess in the medium term, however, the short-term focus here is to configure the Eyecon2 for real-time atline use and prepare samples for analysis on microscope slides as a direct replacement for the microscopemethod. 2018, Innopharma Technology 2 of 10

EquipmentCPS technologyThe Glatt CPS technology is an advanced rotor fluidized bed process. It allows for the manufacture of matrixtype pellets and micropellets with perfect roundness and surface. Direct pelletizing allows for a variety of drugformulations to be realized in extremely low or very high doses.Figure 1: Glatt CPS TechnologyUnlike traditional rotor fluid bed systems, the CPS technology works with a conically tilted rotating disc andadditional devices that allows for a directional particle motion. By means of a characteristic rolling particlemovement and thereby the application of specific centrifugal forces on the arising pellet cores, a defineddensification of the particles can be reached.No starter pellets are required for direct pelletizing with the Glatt CPS technology. The concentrations of theactive ingredients in the pellet matrix can vary from less than 0.1 percent up to a drug load of 90 percent. Round,dense and evenly shaped pellets are achieved, with precisely defined pellet sizes between 100 and 1,500micrometres with an extremely narrow PSD. Microcrystalline cellulose powder is often used as a basic excipient.The pellet formulations can furthermore contain other functional pharmaceutical excipients, such as polymersand disintegrating or solubilizing agents.CPS pellets and micropellets are especially suitable to produce carrier cores for subsequent coating applicationsfor controlled drug release or for taste masking applications.Figure 2:Typical Particle Size Distribution of CPS pellets 2018, Innopharma Technology LtdImage 2: Pellets with 75% drug load from a CPS 3 of 10

Eyecon2 Operating Principles and D-ValuesEyecon2 utilises direct imaging for particle size analysis. In direct imaging particles are illuminated and imagedfrom the same side. The method can be configured for both bench-top and in-line applications, while allowingthe device to remain non-product-contact. This method uses advanced image analysis algorithms to accuratelydetect particle boundaries, and thereby particle sizes, giving real-time information on the process beinganalysed.Eyecon2 provides these real-time results in the format of D-values, histograms, trend graphs and images, and atthe end of every analysis a PDF report summarising the measurement is also generated.A D-value can be thought of as a mass division diameter. It is the diameter which, when all particles in a sampleare arranged in order of ascending mass, divides the total sample mass in the given ratio as demonstrated below.For example, the D10 diameter is the diameter at which 10% of a sample's mass is comprised of smaller particles,and the D50 is the diameter at which 50% of a sample's mass is comprised of smaller particles.The D50 is also known as the "mass median diameter" as it divides the sample equally by mass. 2018, Innopharma Technology 4 of 10

System Configuration for Real-Time At-Line UseA removable sample interface for Eyecon2was designed and configured to allowprocess samples to be measured in asystematic and rapid manner. Eyecon2measurementparameterswereoptimised and a configuration saved forthe type of material being produced bythe CPS process.Image 3: Eyecon2 installation next to a GlattCPS Processor with Pelletization SampleInterfaceProcedureBatch 1 and Batch 2 Parameter settings and Sample PlanBatch 1: The CPS processor was loaded with 800 g pre-wetted MCC 105 granules and set to 1000 rpm, 2 Bar,23.5 C with an initial liquid addition rate of 50 g/min. The liquid addition rate was lowered to 30 g/min afterapproximately 1000 g of liquid addition. Samples were taken after 600 g of liquid addition and approximatelyevery 50 g liquid addition after that point. Each sample was analysed immediately using the microscope methodand then passed for rapid sample preparation and analysis by Eyecon2.Batch 2: The CPS system was set up as per batch 1. Liquid addition was 50 g/min for the first 15 mins and thenreduced to 40 g/min until 2450 g was added. The spray rate was then increased to 50 g/min until the end of therun. Again, once 600 g of liquid had been added samples were taken after approximately every 50 g liquidaddition.Sample PreparationSamples were taken at intervals based on liquidaddition and directly onto commonly availablemicroscope slides (Image 4) which were thenpositioned on the sample interface for immediatemeasurement.Image 4: Samples taken directly to Microscope Slides forimmediate measurement 2018, Innopharma Technology 5 of 10

Sample MeasurementThe Eyecon2 was set up in the process roombeside the CPS process equipment as seen inImage 1 to measure samples at-line in real time.The prepared sample was placed on the sampleinterface. It was decided to measure samplesfrom batch 1 for approx. 20 seconds and samplesfrom Batch 2 for 10 seconds to comparemeasurement time and number of particlesmeasured to optimise time for measurementwith quality of data. The image capture time canbe further optimised on implementation toreduce measurement time if required.Image 5: Real-Time At-Line Configuration of Eyecon2 for CPSAnalysisProcessing of ResultsThroughout the analysis particle information such as D-values, mean diameter and volumetric histogram can beseen in real time along with updated images of the samples measured. The Eyecon2 system generates a standardPDF report on demand for each sample measurement which provides Particle Size Distribution information asstandard including D10, D50 and D90 and provides a volumetric histogram. In addition, example images areprovided. See Appendix 1 for example.In addition, all sample data can be exported in csv format for more detailed analysis and all images can be viewedin browsers and exported for inclusion in reports as required.ResultsThe D-values and mean diameters were plotted vs. mass of liquid added as seen in Figure 3. The particle sizes ateach analysis point are represented by the D-values; D10, D50 & D90, and the mean diameter of the particlesmeasured from the sample. Results for Batch 1 are shown in Figure 3 and trends the particle size change againstliquid addition and demonstrates how the measurements can be used for end-point prediction.Particle Growth during CPS Process - Trial BatchParticle Size Amount Sprayed (g)D10D50D90MeanFigure 3: Particle Growth during a Trial Batch represented by D10, D50, D90 & Mean Diameter 2018, Innopharma Technology 6 of 10

Some sample images captured by the Eyecon2 during particle measurement of a Trial Batch can be seen in Table1. These images are taken from Sample 1, 5, 9, & 15 respectively and demonstrate the presentation of thesample to the Eyecon2, the growth of the particles throughout the process and the issue of particles being outof focus.Table 1: Sample Images from Trial Batch showing progression of process20 Seconds vs 10 Seconds Image Capture PeriodsImage capture time was compared to balance time to generate a result with the quality of the data. Imagecapture of 20 seconds was used for batch 1 and 10 seconds for batch 2. A comparison of the output in numberof images captured and number of particles counted is presented in Table 2.Table 2: Comparison of Analysis TimesAmount Sprayed600 g1250 g1650 g2150 g2750 gNo. of Images Captured 20 sec 10 sec64114116N/A5N/A6No. of Particles Measured 20 sec 10 sec62222203251424901162724N/A671N/A553Overall, the number of particles analysed even at the shortest analysis time for the largest particles is asignificant improvement over the manual method. The manual selection of particles is removed by switchingfrom the manual microscope method to the Eyecon2 method eliminating subjective selection of which particleis measured, as all particles in all images are subjected to image analysis with all particles identified beingincluded in the reported data. Furthermore, the number of particles analysed per sample is significantlyincreased over the traditional method. A review of the table shows the impact of reducing analysis time on thenumber of images captured and therefore the number of particles measured. Generally, the more particles 2018, Innopharma Technology 7 of 10

measured the more representative the sample analysis is of the process. The number of particles measureddecreases as the amount sprayed increases due to the increase in particle size and therefore fewer particles perunit area on the sample slide.It would be possible to reduce the analysis time to 5 seconds but as the trend above shows this would lead tofewer images and therefore less particles being measured, particularly towards the end of the process. It isrecommended that the procedure is optimised for each process to balance the analysis time with the numberof particles counted to achieve the optimal data quality for end-point prediction.Eyecon2 for IPC of CPS and Rotor ProcessesThe CPS process is rapid and particle size is one of the critical quality attributes. Prediction of end-point is keybecause it is important to have a uniform particle with relatively narrow PSD but also important in manyapplications such as taste masking that the particle size remains below a maximum threshold. It is valuable tobe able to rapidly predict the end-point for each formulation to ensure the correct PSD is achieved. Thetraditional IPC relies heavily on operator experience to ‘know’ when the correct PSD is approaching, with thedecision supported by analysis of user selected particles on microscope system. A small number of particles areselected for analysis and recorded as part of the IPC strategy. The selection of particles is largely subjective andcan be identified as an unacceptable risk for IPC in a GMP process.The Eyecon2 coupled with a systematic sample preparation procedure offers a significant improvement to thedevelopment of process parameters and a control strategy that is data-driven and science-based, and thereforemore acceptable for use in-process development and GMP compliant production. The PSD data is generatedduring processing and with optimisation can be used to predict the end-point PSD and therefore when to stopthe process. The most relevant PSD data (Dv50, Dv10, Dv90, mean etc.) will be used for process decisions withthe additional benefit of particle images for each sample point from the process to further aid processunderstanding during development and investigations and for use in technical reports.PSD is a critical quality attribute for the CPS product and process parameters and can be optimised duringdevelopment using real time data from the Eyecon2 leading to a more robust and efficient process withdemonstrably improved IPC strategy. Prediction of process end-point based on PSD will be more data driven andreliable, increasing confidence that later process phases will meet specification. Figure 4 shows a illustrates abatch to batch comparison of PSD change during processing.The process analytical method can also be applied to other rotor processes for process understanding or as partof process control where Particle Size is a Key or Critical Quality Attribute.Figure 4: Multiple Batch Comparison of PSD change during processing 2018, Innopharma Technology 8 of 10

Conclusions/RecommendationsFrom the results of the analysis completed by the Eyecon2 during the evaluation on the CPS process samples itcan be concluded that: The Eyecon2 method could be used to rapidly monitor the increase in particle size with liquid additionduring CPS or other rotor fluid bed batch operations. Samples can be taken directly onto commonlyused microscope slides and immediately measured to provide a rapid update on size during processing. The Eyecon2 system provides real time PSD data to the operator throughout measurement. During andat the end of the session the key PSD data is instantly available allowing an immediate data drivendecision for process control based on PSD. A report for each batch can be generated on demand toattach to a batch record or include in an R&D Technical report. The Standard Report is designed to present data generated in a format and language common toformulation development scientists, process engineers and quality control/assurance and includes D10,D50, D90, mean and median and provides a volumetric histogram. The particle size data and images provide information for a deeper understanding of the impact ofprocess parameter and formulation changes to particle size distribution and therefore productperformance. Once implemented, Eyecon2 data can be used in conjunction with critical processparameter data to correlate with critical quality attributes for process development / improvement /monitoring / control purposes. Set up and operation of the Eyecon2 was straightforward and could be readily incorporated in day today process operations without additional specialist personnel to operate and interpret results.Contact for More InformationEmmet Hogan,Technical Sales Support Specialist, Innopharma TechnologyE: [email protected] 2018, Innopharma Technology 9 of 10

Appendix IPDF Report example 2018, Innopharma Technology 10 of 10

The Eyecon 2 is a non-product contact Process Analytical Technology (PAT) tool used for monitoring particle size in-line or at-line, in real-time. The Eyecon 2 is a direct imaging camera system which provides PSD data and colour images, allowing for a deeper understanding of the key & critical process parameters for a given process/system.