The pharmaceutical industry continues to face challenges developing oral drug products from the large number of low-solubility drug candidates that fall short of 

 exposure targets due to poor absorption. Overcoming this issue requires use of a bioavailability-enabling technology that enhances solubility and/or dissolution rate. Spray-dried dispersions (SDDs) are widely regarded as the technology of choice for improving bioavailability, driven mainly by the technology’s applicability to compounds spanning a diverse physicochemical space and its scalable manufacturing process, offering the flexibility and control needed for an optimized drug product (1–2).

SDDs consist of amorphous drug molecularly dispersed within an excipient matrix—typically, a polymer—and are targeted to provide improved aqueous solubility and thermodynamic driving force for absorption over lower-energy crystalline drug forms. Development of SDD formulations, given their metastable form, is inherently more complex than development of crystalline drug formulations. Success hinges on applying an integrated approach where bioperformance, stability, and manufacturability are considered throughout the development cycle, leading to the simplest, most cost-effective strategy to an optimized SDD formulation.

 outcomes for low-solubility drugs formulated as SDDs. Specifically, 

 preformulation tools and physiologically based pharmacokinetic (PBPK) modeling strategies may help increase understanding of bioperformance mechanisms and make early 

 test selection and design based on drug physicochemical properties, target dose, and species can help advance development.

Belinostat, a poorly soluble drug, offers an example of how drug developers have used 

 performance (3). The chemical structure and physicochemical properties of belinostat are highlighted in 

, respectively. As the small-molecule pipeline contains more solubility-challenged molecules, SDD formulation may become an increasingly important part of the drug development toolkit.

 preformulation tools for compound classification

A preliminary assessment of some of the key absorption hurdles a drug candidate might face 

 is crucial at the outset of a SDD formulation campaign. The goal is to establish a hypothesis about the rate-determining step to 

 absorption, so an initial formulation strategy can be devised regarding selection of the SDD matrix polymer and drug loading.

A preformulation assessment typically starts by predicting and/or measuring the drug’s physicochemical properties and using this information to perform some simple dimensionless-number calculations as part of the Biopharmaceutical Classification System (BCS) and Fraction Absorbed Classification System (FaCS) (4–5). These industry-standard methods are designed to identify potential barriers to absorption based on compound physicochemical properties and projected dose 

. Barriers to absorption generally include drug dissolution rate, solubility, and/or permeability, all of which play a direct role in drug absorption. In the case of belinostat, a fasted beagle dog study was planned at a 50-mg dose to get an initial read on the relative performance of a belinostat SDD formulations compared to that of the crystalline drug form.

For this test plan, the dissolution (Dn), permeation (Pn), and dose numbers (Do) can be calculated using measured aqueous solubility data and the physicochemical properties for crystalline drug. For crystalline belinostat dosed to fasted beagle dogs at a 50-mg dose, the FaCS classification suggests crystalline belinostat will demonstrate solubility-limited absorption. The calculation methods, details of each limiting case, and specific values for belinostat are shown in 

By assessing the rate-limiting case for absorption at the outset of a drug development program, efforts can be focused on the most-promising technology (e.g., SDD) while weighing the associated benefits and risks. In the case of belinostat, it was clear that if absorption were to be improved, the best path forward would likely be via SDD formulation, because both aqueous solubility and dissolution rate would increase compared to crystalline drug, likely leading to improved absorption and bioavailability.

 tools for formulation design and bioperformance evaluation

 problem statement identification tools, developers may then select and design 

 tests based on three main criteria: the physicochemical properties of the compound, the target dose and species, and the key performance attributes of the formulation based on the (hypothesized) rate-limiting step to absorption and target pharmacokinetic profile.

For belinostat, the key tests employed during SDD development were, first, amorphous solubility/polymer screening and, second, a gastric-to-intestinal transfer test to determine dissolution performance. Fiber-optic ultraviolet (UV) probes are used for these tests, which are described in 

. Data from the tests are used as inputs into a pharmacokinetic model for 

. The test for amorphous solubility/polymer screening is often the first and most important 

 assessment before formulation screening. Amorphous solubility is measured via solvent-shift UV assay with fiber-optic UV probe detection, where drug is titrated from a stock solution in organic solvent into a dissolution medium of interest (with and without predissolved polymer). As stock solution is titrated, wavelengths are monitored beyond the UV cutoff of the drug in an attempt to observe any light scattering that occurs once liquid-liquid phase separation is reached, the key indicator of the amorphous solubility.

Drug amorphous solubility: Provides an initial assessment of the solubility benefit of the amorphous form relative to crystalline drug, which can be fed back into the FaCS calculation to reassess, if necessary, the rate-limiting parameter for absorption of the amorphous drug form.

Polymer sustainment: Measures the ability of each polymer to sustain supersaturated drug concentrations (i.e., sustainment), a critical metric for success for most SDD formulations.

Drug/polymer interaction: Identifies unique drug behavior in the presence of each polymer (e.g., polymers may suppress the drug amorphous solubility).

For belinostat, this test showed the amorphous solubility was substantially higher than the crystalline aqueous solubility (

). Sustainment of supersaturation was excellent for the three polymers that were evaluated: polyvinylpyrrolidone (PVP K30), PVP vinyl acetate (PVP VA64), and the M grade of hydroxypropyl methylcellulose acetate succinate (HPMCAS-M). Interestingly, the amorphous solubility of belinostat was significantly altered by the presence of polymer and was specific to each polymer type. This result suggests that polymer selection for formulation screening would hinge more on the attenuation effect of the amorphous solubility than on sustainment of supersaturation. It also suggests that each formulation has a different ceiling with respect to the overall solubility enhancement relative to crystalline drug.

Lastly, recalculating the FaCS classification for amorphous belinostat suggests that belinostat SDDs will be dissolution-limited rather than solubility-limited, driven by the formulations’ substantial solubility enhancement over crystalline belinostat. As such, maximizing the rate and extent of SDD dissolution should be a key focus area.

. Dissolution testing using fiber-optic UV probes is a quintessential part of the 

 bioperformance toolkit for many drug developers and manufacturers. Fiber-optic probes afford the opportunity to perform real-time data collection and analysis with high time resolution and minimal effort compared with conventional manual sampling methods. They also provide the opportunity to perform orthogonal characterization concurrent with the experiment to establish a mechanistic understanding of formulation performance with respect to dissolution, precipitation, and drug speciation.

Given the results of the amorphous solubility/polymer screening test, prototype SDD formulations were manufactured at a 25% (w/w) belinostat drug loading with PVP VA64, PVP K30, and HPMCAS-M. These formulations were then tested in a gastric-to-intestinal transfer dissolution test. Experimental conditions were chosen based upon the amorphous solubility results and the nominal dose/volume for a fasted beagle dog to optimize biorelevant conditions. Test conditions, method descriptions, and results are shown in 

The dissolution data aligned well with the amorphous solubility results for the SDDs in simulated gastric fluid (SGF) (nonsink dose) with the exception of the HPMCAS-M SDD. HPMCAS-M is an enteric polymer with low solubility at pH values below 5.5, resulting in limited dissolution of belinostat from the SDD particle in SGF. In simulated intestinal fluid (SIF) (sink dose for all SDDs), the results suggested the following rank by performance: PVP K30 > HPMCAS-M > PVP VA64. The PVP K30 SDD was clearly superior in SGF, providing the highest dissolution rate and dissolution extent of all SDDs.

Pharmacokinetic modeling for hypothesis testing and 

 evaluation is to build a pharmacokinetic model incorporating key 

 performance. The goal is to continue the hypothesis-driven formulation strategy by setting expectations 

 that are aligned with the formulator’s understanding of drug physicochemical properties, 

 study design, and how each formulation performs 

 with respect to dissolution, precipitation, and drug speciation.

For belinostat, this includes incorporating physicochemical properties, amorphous solubility data with unique solubility inputs for each SDD, as well as dissolution kinetics for each SDD in both the stomach (ca. pH 2) and the intestine (ca. pH 6.5). Pharmacokinetic parameters were established using intravenous data generated in beagle dogs, and the resulting simulations were compared to the observed 

Results suggest that the simulation using amorphous solubility and dissolution data did an excellent job predicting the 

 outcome. Additional model parameters for sensitivity analysis (not shown) helped elucidate the key metric for 

 success for these formulations: dissolution rate and extent of dissolution achieved in the stomach before gastric emptying into the intestine for absorption.

SDD technology is a unique and broadly applicable platform for improving the oral bioavailability of poorly soluble drugs. However, successful SDD development requires a deep scientific understanding of amorphous systems and their behaviors 

 tools when developing SDD formulations saves time and resources during development by improving understanding of the key barriers to absorption, as does possessing the tools to accurately evaluate these types of formulations 

The belinostat SDD case study demonstrated the use of these tools and concepts, ultimately leading to an optimized outcome and successful 

 testing. As low-solubility and low-bioavailability molecules grow more prevalent in small-molecule drug development, having access to best-in-class SDD manufacturing capabilities may become more important for pharma and biotech innovators. Working with external partners may help provide access to these capabilities, particularly for smaller or emerging firms that often lack sophisticated manufacturing capabilities in-house.

 4(3) 133–142 (2009).
2. A. Singh and G. Van den Mooter, 

 100 27–50 (2016).
3. A. Stewart, et al., 

 Sci 108(1) 326–336 (2019).
4. K. Sugano and K. Terada, 

 104(9) 2777–88 (2015).
5. G.L. Amidon, et al., 

 12(3) 413–420 (1995).
6. E.R. Grignard, et al

 99 193–201 (2017).
7. M. Arndt et al, 

 84(3) 633–641 (2013).
8. P.K. Gupta and J.R. Robinson, 

Aaron Stewart is associate principal scientist at Lonza Pharma and Biotech, Bend, OR.

Solid Dosage Drug Development and Manufacturing

When referring to this article, please cite it as A. Stewart, “Advancing Spray-Dried Dispersion Formulation Development,” 

Articles

Development

In-vitro and in-silico tools can help predict in-vivo outcomes for low-solubility drugs formulated as spray-dried dispersions. 

Advancing Spray-Dried Dispersion Formulation Development

Demanding regulations and tight quality tolerances drive the need for accurate analytical information in real time to ensure manufactured drugs are safe and effective. Process analytical technology (PAT) must deliver fast and data-rich analysis so that manufacturers can accurately measure critical process parameters (CPPs) that affect critical quality attributes (CQAs). PAT is crucial to the process of manufacturing drugs in compliance with regulations.

PAT uses established spectroscopy technology and advanced statistical computer modeling to deliver information about the manufacturing process. Optimal system characteristics include fast, easy access to analytical data to address production problems with minimal downtime; a nondestructive process so drug product is not wasted in quality testing; reliability; and low maintenance.

Raman spectroscopy measures the light-matter interactions (the scattering of photons due to bond vibrations in a molecule) via a laser and detector. The vibrations generate a Raman spectrum, a unique molecular fingerprint of the analyzed substance, providing both quantitative and qualitative molecular compositional information. The chemometric models produced by a Raman system by collecting and preprocessing molecular compositional data are linear and repeatable.

Raman technology has been used in pharmaceutical manufacturing for more than 10 years to analyze solids, powders, liquids, gels, gasses, and heterogeneous samples, and is recognized as in-line PAT tool to measure CPPs.

Because Raman systems use a laser to excite the molecules, measurement times range from milliseconds to seconds and do not require sample prep steps or separations. Hardware advances have improved reliability, ease of use, data quality, signal strength, communications, and the ability to interface with multivariate analysis software packages.

PAT is recognized as an important tool for monitoring pharmaceutical manufacturing processes, and Raman spectroscopy has emerged as a fast, effective, and versatile PAT real-time analyzer for multiple stages of the manufacturing process. 

illustrates the typical unit operations in pharmaceutical manufacturing.

Raman spectroscopy has been applied as a multidimensional PAT sensor for real-time monitoring of CQAs and real-time release of products. Further, Raman spectroscopy has been used in continuous secondary manufacturing and small-volume continuous manufacturing of APIs.

Raman spectroscopy provides a molecular fingerprint of incoming excipients and APIs. As a result, Raman is used by many major pharmaceutical companies for verification and identification. Two critical Raman attributes that support verification and identification are specificity and sensitivity. Raman can identify different polymorphic forms where significant differences are seen in the collected Raman spectrum’s low-frequency region. Raman also identifies material adulterations, whether accidental or intentional. In 

, the Raman component results of Eno, an over-the-counter antacid composed of sodium carbonate, sodium bicarbonate, and citric acid, are shown.

Measurements for analysis are made either by direct contact using a handheld fiberoptic probe or a contactless probe, which analyzes a sample through a glass or plastic container. Once the Raman spectrum is collected, it is compared to a commercially available or custom-built library. The output is a simple indication of pass/fail. If the material passes the identification/verification stage, it can be released into the system. If the material fails, a sample is taken for further laboratory analysis, and that lot could ultimately be rejected.

On-line analysis of blend uniformity by Raman spectroscopy provides more benefits than traditional sample extraction (thief sampling) and off-line analysis. On-line Raman spectroscopy obtains a large number of spectra, which generates data-rich blend profiles during the blending process, with spectral collection rates collected in approximately two seconds.

A small solid-state Raman spectrometer connected to a fiberoptic probe can be installed on various blending systems such as V-blenders or intermediate bulk containers (IBC). For IBC implementation, the spectrometer must rotate with the IBC, so the instrument must be small, battery-operated, and able to transmit spectra in real-time wirelessly to a remote computer.

A model-free statistical approach monitors the blend uniformity by calculating a spectral region’s moving block standard deviation specific to the API (

). Once the standard deviation reaches the threshold required for a period of time, the blend is deemed uniform and is stopped, and the batch is released. This use of real-time Raman spectroscopy can reduce batch cycle times and ensure product quality and consistency.

In the manufacture of pharmaceutical tablets, Raman spectroscopic methods can provide a real-time, nondestructive measurement of the API concentration to ensure product quality and uniformity and enable real-time release of a batch.

A port on the feed frame or a sapphire window allows either contact or contactless material measurement by Raman spectrometers before tablet compression. The Raman probe is a crucial component in obtaining high-quality spectra.

For the contact measurement, the focusing of the Raman excitation beam or probe cleaning is not necessary if using a touch probe design that is self-cleaning in a moving powder stream.

A contact-free interface can be used to measure through a sapphire window; a standoff distance of approximately 10 mm will ensure that the sapphire window has little or no influence on the Raman spectra.

In conventional batch analysis, operators collect the required number of tablet cores throughout a batch, based upon the standard operating procedure for sampling each batch. The sample set is sent for a high performance liquid chromatography analysis on each tablet, a process that can result in a batch being held for several hours, if not an entire day.

By using Raman spectroscopy, each tablet core is analyzed using a contactless probe and closed tablet holder, generating analysis results in seconds. With this method, the entire sample set can be analyzed in less than one hour before releasing the batch.

In proprietary tests, Raman spectroscopy has been shown to lower the limit-of-detection due to excellent specificity and sensitivity when compared to other spectroscopic technology, particularly for formulations with low API loads.

Tablet film coating is a standard pharmaceutical process in which various coating formulations are applied to tablet cores. To ensure the product’s efficacy and safety, along with the desired dissolution profile, consistent quality of the coating is required. A CQA of the film coating is the film thickness, which can be monitored in real-time by Raman spectroscopy.

In unpublished customer application data, a high-volume, contactless probe with a beam diameter of 3.5 mm and a standoff distance of 100 mm has been used to measure film thickness on a perforated drum coating machine (

). Raman spectra are collected every few seconds and analyzed using a multivariate data analysis (MVDA) model.

Real-time coating thickness results can be verified visually and sent to a control system via industry-standard protocols such as OPC UA (Unified Architecture) or DA (Data Access), Modbus, Profibus, or 4-20mA outputs. The control system optimizes the process by adjusting the coating spray rate or drum speed to ensure that the final product meets the coating thickness specification. This step will ensure that no out-of-specification tablets are produced while improving the process efficiency by potentially reducing the batch coating time and ancillary testing.

Pharmaceutical manufacturers understand the value multidimensional PAT technology offers in ensuring continued high-paced operations. As a PAT sensor, Raman can deliver accurate and data-rich compositional analysis in approximately two seconds—depending on the application—in a nondestructive process minimal sample preparation.

Accessible and portable Raman systems can be used throughout the pharmaceutical manufacturing process from mixing to blending. Matching the appropriate system to the pharmaceutical process is key to maximizing the capabilities of Raman technology.

Brian Marquardt, PhD, is founder and CEO of MarqMetrix, Inc., and director of the Center for Process Analysis and Control (CPAC) at the University of Washington.

When referring to this article, please cite it as B. Marquardt, “Raman Detects Pharma Processing Fingerprints,” 

Manufacturing

Raman spectroscopy delivers real time, non-destructive process insights for critical process parameters to monitor pharmaceutical manufacturing.

Raman Detects Pharma Processing Fingerprints

Powder mixing, a key step in the manufacturing of solid dosage pharmaceuticals, is more challenging than mixing liquids. While it can be relatively easy to move liquids around by optimizing pumping and piping systems, bulk solids do not flow easily, making materials transfer more difficult. Processing solids depends on controlling and optimizing particle-size distribution and dynamic flow properties such as flowability and compressibility. These properties are inherent material characteristics and determined experimentally. In the past few years, pharmaceutical manufacturers have adopted computer-based simulation approaches to reduce the cost and improve the results of technology transfer and equipment selection, and to improve the overall solids mixing process itself (

In technology transfer, a product moves from the laboratory to a pilot plant or the manufacturing floor, or from one equipment scale or type to another. Previously, this transition was a weak link in drug development. R&D professionals used to speak of throwing processes “over the wall” to manufacturing. Traditionally, the most common approach to technology transfer has relied primarily on a reiterated cycle of testing, trying, learning, then retesting, a costly and time-consuming approach. Such testing was done manually, using bench-scale testing methods to mimic manufacturing operations (e.g., blending, solids dissolution, and filling).

However, manual techniques often failed to ensure that processes scaled well to a production setting. In addition, the processes could be tedious, error-prone, expensive, and time consuming. Often, a large amount of design-of-experiment (DoE) data are needed to determine the conditions that will maximize product yield, another aspect of traditional technology transfer that adds complexity, cost, and time. 

Simulation tools can be used to develop a roadmap for technology transfer involving changes in materials, scale, and process variability based on a fundamental understanding of the processes and science involved (1). Simulation can benefit internal transfers or those to a contract manufacturing organization (CMO) or contract development and manufacturing organization (CDMO). Digital tools are increasingly being used to improve mixing. The mixing equipment used at one site such as R&D lab may be different from the equipment that will be used at the site where the process will be transferred such as a CDMO’s manufacturing site. This particular tool focuses on differences between impeller-generated flow patterns in two mixing vessels at the different sites, as illustrated in  

. It uses digital solutions-based analysis to characterize mixing behavior and extract key performance parameters for the process, allowing the technology transfer team to achieve parity between parameters in the two vessels by adjusting individual process parameters and/or equipment configuration. This approach minimizes expensive testing and reduces development time and cost.

Digital tools can also be used to determine the suitability of a specific type of equipment for the given process and product. For example, skid-mounted systems are designed to offer portability, flexibility, and quick changeover, and to allow the same equipment to be used in multiple applications. Mixing and blending operations are commonly placed on a skid.In deploying a skid-based mixing system, the following questions must be considered:

Does the equipment have sufficient horsepower to run the process?

Does the vessel configuration provide adequate mixing?

How can proper mixing in the vessel be assured?

Is the batch over-mixed, under-mixed, or the mixing is just right?

For solids suspension, is there settling of solids in the vessel that can result in product nonuniformity?

Is any air, which can affect efficacy and shelf life, entrapped in the product?

How can solid suspension be balanced with air entrapment? Aggressive mixing ensures solid suspension but entrains too much air, while gentle mixing does not entrain air but may result in settling of solids.

What is the best location in the vessel to sample the product for quality assurance?

How can the process be run most efficiently?

Is there opportunity to improve throughput and
reduce cost?

In a traditional setting, these questions would be answered during equipment procurement or process development and deployment stages. However, in a flexible manufacturing environment, these questions arise more frequently and must be answered quickly. For these situations, digital solutions-based analysis can be coupled with selective rheological or bench-scale testing to provide a rapid way to determine whether equipment will be suitable for a specific application. This approach minimizes risk and waste, allowing informed decisions related to suitability of equipment to be made well in advance of deployment.

For example, consider a skid-based system used to manufacture a suspension product. Sufficient agitation is necessary to prevent settling of particulates from the suspension. At the time of skid deployment, it is crucial to measure the impeller’s revolutions per minute (rpm) for the process. However, excessive agitation will lead to air entrapment and affect the efficacy of the product. Optimal rpm is therefore necessary to ensure suspension uniformity while minimizing air entrapment.

Digital analysis can be used to determine the impact of rpm on particle concentration in the vessel. This analysis can also be applied to extract key parameters that result in air entrapment. 

 shows how modeling can be used to optimize the balance of required properties.

For powder/granular flow analysis, the digital simulation technique is selected depending on the powder flow regime. For situations where powder/granular flow is rapid, the predictive methods developed for fluid flow apply. In cases where friction dominates, both discrete element methods as well as fluid flow techniques apply. Use of these methods is illustrated in the following example.

For tablet and capsule manufacturing process, binders, flavoring agents, colorant, disintegrants, anti-adherent agents, glidants, and lubricants must all be added to the primary excipient and API and mixed. The application of digital simulations to this type of mixing application is shown in 

. In this case, the mixing of two granular ingredients is simulated using discrete element methods. The interactions between the granular particles and the enclosing vessel are simulated based on the fundamental laws of physics. In this example, the equipment is not optimal for mixing and the simulation shows the segregation of ingredients. Simulations like this can be applied, not only to select the best equipment, but to optimize process conditions to achieve optimal mixing.

Digital simulation-based analysis coupled with selective bench-scale testing provide a rapid means to determine suitability of equipment for different applications. The use of predictive analysis minimizes risk and waste so informed decisions related to suitability of equipment can be made prior to deployment. Other processes where digital simulations coupled with selective customized testing have been leveraged for process development and technology transfer include lyophilization, spray drying. and powder-filling systems. The heat, mass, and momentum transfer associated with particle transport during spray drying can be simulated with a great deal of accuracy providing high fidelity digital models that are leveraged to optimize and troubleshoot the process and associated equipment. Similarly, conveying of powder during filling can be evaluated to prevent issues associated with fillers such as dust generation and deposition of powder over sealing surfaces of packages.

As the industry moves toward Pharma 4.0, digitization, data analytics, machine learning, and artificial intelligence will increasingly be used to improve productivity and quality while reducing costs. It must be recognized, however, that this cannot be achieved through data alone; a hybrid approach leveraging physics and data science is necessary to realize the benefits. Solid-dosage forms are manufactured using ingredients in powder form. Digital simulations provide a means to quantify critical process parameters (CPP). Powder materials exhibit complex behaviors and it is not possible to relate the critical quality attributes (CQA) of the final product to CPP based purely on computational methods. Customized testing is required to make the connection between input parameters and CQA.

In the near future, the industry may see increased use of a hybrid approach that combines physics-based models, customized test data, and artificial intelligence algorithms to generate surrogate models that relate ingredients, process equipment attributes, and process conditions to CPP and CQA. At this point, however, digital simulation tools continue to gain ground in solid dosage form development and process optimization.


Supplement: Solid Dosage Drug Development and Manufacturing
April 2021
Pages: s19–s21

When referring to this article, please cite it as H. Prodal and C. A. Haynes, “Harnessing Predictive Simulation to Improve Mixing” 

Pharmaceutical Technology Solid Dosage Drug Development and Manufacturing

Predictive computational methods are finding wider use in pharmaceutical solid dosage development, particularly in mixing and blending, to improve tech transfer and equipment selection, and speed the acquisition of process understanding.

Harnessing Predictive  Simulation to Improve Mixing

Process analytical technology (PAT) tools collect process data that can be used for real-time process control and to increase process understanding. 

 interviewed Eamonn Garry, operations director at Optimal Industrial Automation, and Paul Gillham, innovations director at Optimal Industrial Technologies, about implementing PAT in oral solid dose (OSD) manufacturing.

 PAT provides OSD manufacturers with a data-driven, in-depth process understanding that supports accurate and responsive in-line control across all production stages, from raw material input up to the delivery of end products. Thanks to the actionable insight gained from these data, it is possible to create an intelligent as well as flexible manufacturing line that is able to address variability and meet market demands with high-quality products. PAT is therefore essential for any business seeking to implement quality by design (QbD), lean, and smart manufacturing applications.

By offering a holistic approach to manufacturing, PAT offers multiple advantages to pharmaceutical producers. First, this system is crucial for the transition from traditional quality control methods to more comprehensive and efficient quality assurance strategies. As a result, OSD producers can improve the consistency of end-product quality whilst considerably reducing the volume of off-spec materials, rejects, and reworks. Furthermore, the advanced process understanding generated by PAT enables manufacturers to continuously increase the quality of their end products by fine-tuning operating conditions.

Moreover, the information on quality that is continuously generated, analyzed, and stored offers full data traceability, which facilitates quality auditing activities to prove regulatory compliance. In practice, this provides the foundation for real-time release testing (RTRT). All these quality-oriented features also help manufacturers increase product safety, thus minimizing—if not eliminating—the risk of drug recalls.

A shared PAT infrastructure can also enhance collaboration between manufacturing facilities within an organization, improving end-product consistency and maintaining quality across the company.

In addition to increased quality, pharmaceutical companies can also benefit from more energy-efficient and resource-efficient processes. Not only do fewer off-specs and reworks result in lower waste generation and material usage, but optimum process conditions also avoid overprocessing. As a result, OSD producers can reduce their running costs, leading to substantial savings. In particular, PAT can increase operating margins by 6% for mid-sized pharmaceutical companies manufacturing small-market branded, generic, and over-the-counter OSD forms (1).

Furthermore, the reduction of reworks and overprocessing, coupled with the elimination of downtime associated with off-line quality control, play a key role in shortening production times. This [faster production], in turn, increases productivity and speed to market. For new OSD formulations, producers can take full advantage of a product’s patent life.

Manufacturers wanting to fully optimize their operations can also leverage PAT to facilitate the transition from batch to continuous processing, which can achieve even more productivity gains. For example, Vertex reported how this change in production has helped the company manufacture medicines from APIs to commercial-ready tablets in a few days as opposed to 4–8 weeks with a traditional batch process (2).

Finally, PAT-driven applications use the power of data, in line with Industry 4.0 or Pharma 4.0 concepts. Consequently, businesses have a key tool that enables them to continuously drive process improvements and intensification as well as to increase their flexibility in adapting to variability in market demands. OSD manufacturers can therefore address current market requirements while being ready to meet future needs.

 What are the challenges for installing such systems in existing lines (retrofitting)? Is this possible/practical? Or is it best for new lines?

 One of the biggest benefits of PAT is that both new and existing manufacturing lines can implement and profit from it, whether they run batch or continuous processes. Therefore, businesses adopting a PAT-driven approach do not need to build new facilities. In fact, they can leverage this framework to retrofit their plants, improving their operations.

Whether PAT is applied to grassroots or brownfield plants, it is important to seamlessly integrate the technology with the facility’s control system. Therefore, it is crucial to design and validate a well interconnected setup.

The implementation of PAT on existing manufacturing lines requires the development of a PAT environment that can be adapted to the existing infrastructure. It may also be necessary to adapt the pre-existing control system to a quality-centric control strategy as directed and orchestrated by a PAT system.

In new facilities, it is beneficial to develop the control and PAT aspects simultaneously, so that a comprehensive solution can be created that is tailored to the application. Ultimately, the creation of a PAT-driven OSD line should follow the same principles of QbD that are applied for process and quality monitoring, where quality is built into the solution from the outset.

In both cases, companies should rely on a PAT specialist, especially if they do not have expert in-house teams. The right specialist should have extensive experience in the pharmaceutical sector and a proven track record in developing customized solutions for new and existing plants. This [expertise] will ensure the creation of a suitable framework that addresses the specific needs of the OSD manufacturing plant. By collaborating with such an expert, businesses can make sure they are using a high-quality system that maximizes the benefits of PAT-driven manufacturing.

 When thinking about using PAT for automating/controlling an OSD manufacturing line, what are some of the things to consider?

 The most important element that should be considered in automated, PAT-driven OSD manufacturing lines is certainly interoperability. PAT systems are highly likely to interface with multiple systems and devices, including instruments and analyzers, programmable logic controllers (PLCs), supervisory control and data acquisition (SCADA) systems, historians and, potentially, distributed control systems (DCS). Therefore, it is crucial to select a PAT knowledge management platform for process orchestration and closed-loop control that can communicate bi-directionally with multiple automation, device, and system vendors.

For the same reasons, it is highly beneficial to select an independent PAT platform and system integrator specialist that has experience with automation products from different manufacturers, as well as in-depth understanding of PAT and how to deploy the technology effectively. By utilizing this knowledge, the specialist can select the most suitable components for an intended application as well as effectively configure devices on new or existing networks.

Another aspect that should be considered is the structure of the PAT knowledge management software used. In particular, a modular solution should be favored, as it allows specialists to conduct the installation, validation, and testing of PAT setups in different stages, which can then be unified. This ability is especially important in continuous manufacturing lines, where everything should operate in sync. In these situations, having the flexibility to develop the automation and PAT strategy for each unit operation separately enables the simplified, ‘standalone’ testing of each component before each one is successfully incorporated into a larger, comprehensive automation and PAT framework for integrated testing and subsequent use.

In addition, modularity not only simplifies the creation of a suitable PAT-led manufacturing approach, but also helps manufacturers to adopt the framework in a staged PAT strategy. Such a strategy allows progressive benefits to be gained as the technology is rolled out along the processing line. In effect, OSD manufacturers adopting this manufacturing concept can begin with a simple, smaller-scale project that would enable the business to explore PAT and the possible gains while getting accustomed to the technology.

 What types of maintenance must be performed when using these systems?

 Another key advantage of PAT is that once a suitable system has been correctly set up, the framework can operate near continuously. A well-designed PAT system can also be adjusted to support expansions, modifications, or the repurposing of existing lines and units. In addition, it can support its own optimization by using increasingly accurate and representative predictive models as more process data is analyzed. For this reason, a PAT approach embraces continuous improvement.

To realize these opportunities, a modular PAT infrastructure can be highly beneficial. In particular, it facilitates a highly flexible, extensible, and scalable PAT approach. As a result, its reprogramming or development, validation, and subsequent addition to the complete infrastructure can be simplified and streamlined.

 2, 38–50 (2007).
2. Manufacturing Chemist, “Talking continuous: Q&A with Patricia Hurter from Vertex,” 


Supplement: Solid Dosage Drug Development and Manufacturing
April 2021
Pages: s26–s28

When referring to this article, please cite it as J. Markarian, “Using PAT in OSD Manufacturing,” Solid Dosage Drug Development and Manufacturing Supplement (April 2021).

Manufacturers of oral solid dose pharmaceutical formulations need PAT tools to reduce a product’s time-to-market and increase profitability. 

Using PAT in OSD Manufacturing 

During a hands-on seminar held in July 2019 in Leverkusen, Germany, the authors compared the performance of five medium-sized rotary tablet presses to see how well they handle formulations that are prone to causing poor flowability, overlubrication, and capping. The equipment vendors IMA, Kikusui, Korsch, Romaco Kilian, and Syntegon (formerly Bosch), whose presses were used in the tests, were allowed to adapt filling system configurations for each formulation, without changing punches or dies.In order to ensure anonymity, no actual names were assigned to the results; the presses are identified as RP 1, RP 2, RP 3, RP 4, and RP 5.

The following trials were performed for each of the formulations at the same theoretical dwell time: A force-hardness profile was generated to compare the tableting behavior of the various presses. The next step was either to minimize the standard deviation of tablet weight (s

), or to investigate the influence of precompaction pressure. The final challenge was to maximize output while staying in spec.

provide overviews of the relevant properties of the participating rotary tablet presses and the used fill shoe configurations. All rotary presses used the same standard Euronorm EuB tooling with a punch tip diameter of 8 mm and a radius of curvature of 12 mm. EuBB dies with the same tapering dimensions were used.

To test the rotary tablet presses, the formulations consisted mainly of excipients that are seldom used for direct tableting, or are used in smaller quantities. All of the formulations were blended in a drum hoop mixer (J. Engelsmann AG, Germany).

Formulation 1 contained of 74% GranuLac 200 (lactose, Meggle), 24.5% Tablettose 70 (lactose, Meggle), and 1.5% LIGAMED MF-2-V (magnesium stearate, Peter Greven). This formulation is characterized by its poor flowability (

), because of the large amount of GranuLac 200. GranuLac 200 has an average particle size of 30 μm, and is typically used in wet or dry granulation processes. To improve the flowability, some Tablettose 70, an agglomerated lactose with an average particle size of 190 μm, was added. Despite the difference in particle size, the flowability of this blend was still sufficiently poor so that segregation would not likely occur.

 Formulation 2 was compounded of 84% UNI-PUR WG 220 (pregelatinized corn starch, Ingredion), 14.5% VIVAPUR 105 (microcrystalline cellulose, JRS), and 1.5% LIGAMED MF-2-V (magnesium stearate, Peter Greven). Formulation 2 was sensitive to densification speed and prone to overlubrication. The latter was tested before the seminar began. Increasing mixing times and intensities resulted in a decrease in tablet strength. The flowability of UNI-PUR was reduced by adding a poorly flowing microcrystalline cellulose with an average particle size of 15 μm. This forced the tablet presses to use a higher impeller speed and mixing intensity.

 For Formulation 3, 78% of the brittle TRI-CAFOS 500, a spray-dried tribasic calcium phosphate (Budenheim), which is mainly used as a disintegration enhancer in concentrations of 10–30% (1), was mixed with 20% highly elastic corn starch MAIZE STARCH BP (Ingredion) and 2% LIGAMED MF-2-V. The goal was to obtain a formulation that showed capping at higher tableting speeds.

To compare the density of the powder during filling with the bulk and tap density, the filling density of the powder was calculated from the filling height, the dimensions of the die, and the tablet weight. Tablet mass, height (thickness), diameter (D), and crushing force (CF) were determined with a semi-automated tablet hardness tester (ST-50, Sotax). The tensile strength (TS) of the tablets was calculated according to Francke (2) using the formula for equivalent tablet height (h

In order to characterize tableting properties, tensile strength was plotted against compaction pressure or tablet density. These plots included the single values of 10 tablets per compaction pressure level as well as their average value. The standard deviation of tablet weight (s

) is based on the weight of 50 tablets. The maximum allowed s

 was determined by modifying past approaches (3). The experimental setup for all three formulations was quite similar. At first, extended force-hardness profiles were made at compaction pressures of 50 MPa to 300 MPa without precompaction pressure.

The second task was either to minimize the standard deviation of tablet weight (as with Formulations 1 and 2), or to examine the influence of the precompaction force (Formulations 2 and 3). Finally, the presses were challenged to maximize the output without allowing a standard deviation of tablet weight above 2% or a tablet crushing force of less than 80 N (Formulation 1), 40 N (Formulation 2), or 60 N (Formulation 3).

Problem formulation 1: Poorly flowing material. Figure 1 

shows the extended force-hardness profiles for all five rotary tablet presses. In 

, tablet tensile strength is plotted as a function of compression pressure and in 

 as a function of tablet density. Three presses had to be used with a small precompaction force of 1.3 – 2.1 kN (precompaction pressure: 25–40 MPa) to prevent damage to the main compression roll.

, the tensile strength versus compaction pressure profiles were not identical for the various presses. This is not unexpected, because this plot is based on the machine parameter compaction pressure. Any differences that were seen could have resulted from differences in the following:

Densification process of the powder bed height (e.g., due to small differences in the overall densification profiles).

Die fill process, due to specific machine vibrations.

Force measurement systems, which are calibrated statically, without proof that the statically obtained calibration data are also valid during the dynamic tableting process.

, which shows tensile strength versus tablet density, is based on tablet properties only, and is independent of machine parameters. Data from four of the five presses coincided. Only RP 1’s results differed slightly from those of the other presses: At the same tablet density, a lower tensile strength was obtained. This might be due to differences in the rearrangement phase of the powder particles. Additionally, this might be explained by mixing intensity within the fill shoe. 

The latter hypothesis is supported by the results from 

, showing the powder densities during filling, calculated as percentage of the tab density. RP 1 was the only tablet press that densified the powder bed during filling.

shows the standard deviations of tablet weight (s

) at two different dwell times: 9 ms and 6.5 ms. All rotary presses were able to produce tablets with an srel of 1.5% or less, which is sufficient for a tablet weight of 240 mg.

Interestingly, RP 1 achieved the lowest s

 of 0.54% and 0.65% for the higher tableting speed. This seems to correlate with die-fill density, which is equal to tap density, and therefore higher than it is for the other tablet presses. For RP 2, RP 3 and RP 4, the die fill density nearly equaled the bulk density of the powder, while, for RP 5, the obtained die-fill density was even slightly lower than the bulk density. These differences are most likely due to differences among the fill shoe devices in the various presses. Higher powder density was achieved within the RP 1 fill shoe than within the fill shoes of the RP 2, RP 3, and RP 4 presses. At the same time, the powder in the RP 5 fill shoe appeared to have fluidized, slightly.

For Formulation 1, densification of the powder bed within the fill shoe is beneficial for achieving the standard deviation of tablet weight. In this case, the small loss in binding capacity (

) was irrelevant because, with all presses, the minimally required tablet strength was achieved. In 

, the results of the challenge are summarized. Upon an increase of the tableting speed, s

became larger for RP 1, RP 2, and RP 3, while it stayed constant for RP 4 and RP 5 (compare to 

For the other presses, a slight increase of s

 was observed, but all presses achieved an s

 of less than the requested 2%, surprising results, given the poor flowability of Formulation 1. The target crushing force of 80–100 N was achieved for all presses, and the maximum outputs of all presses are comparable. Relating these outputs to the maximum possible levels would have been an interesting exercise but would have allowed the identities of the individual presses to be identified. The range was 83–100% of the maximally possible output. From these data, it can be concluded that even very poorly flowing material can be processed with modern rotary tablet presses at high speed and an appropriate standard deviation of tablet weight.

Problem formulation 2: Tendency to overlubricate.

 shows the extended force-hardness profiles of Formulation 2. As with Formulation 1, the compaction pressure versus tensile strength profiles (

) were not identical, and deviated more than the tensile strength versus density profiles (

). The explanation for the larger differences is the same as it is for Formulation 1. Differences between the tensile strength versus density profiles, if any, are practically irrelevant. With RP 3, the lowest tablet density and therefore the lowest tablet strength was obtained, most likely because this was one of the two presses that operated at zero precompaction force.

 summarizes the tensile strength values obtained, as well as the applied pre- and main compaction forces and the speed of the fill shoe impeller. 

All presses were able to produce tablets that provided the requested crushing force of at least 40 N. By applying a larger main compaction pressure, larger crushing forces were also achieved. Additionally, applying a precompaction pressure in general resulted in stronger tablets. When increasing the tableting speed, the impeller speed had to be increased as well. Nevertheless, no relevant overlubrication could be observed. The target crushing force of 40 N was still achieved with all five rotary presses.

Differences between the presses could be observed with respect to the impeller speed used, which is much higher for RP 1, RP 2, and RP 3 than for RP 4 and RP 5. The risk of overlubrication by the impellers distributing the lubricant more intensively within the powder bed is therefore smaller for RP 4 and RP 5. In addition, differences in the applied compaction pressure could be observed. Whereas a much higher pressure was applied for achieving the requested tablet strength with RP 1 and RP 3, the rotary presses RP 2, RP 4, and RP 5 could manage with less.

Whether these differences in force are caused by a reduction of tablet strength due to intensive stirring in the fill shoe or by the speed sensitivity of the formulation could not be concluded from this experimental setup. 

 compares the tablet crushing force with the achieved output in tablets per hour, and shows the variations in tablet weight that were obtained. 

All of the presses tested fulfilled the required values: 40 N crushing force and the s

Tests evaluated commercial rotary tablet presses to see how effectively they deal with problems such as poor flow, overlubrication, and capping.

Pharmaceutical Technology-04-01-2021

Advancing Spray-Dried Dispersion Formulation Development

Raman Detects Pharma Processing Fingerprints

Harnessing Predictive Simulation to Improve Mixing

Using PAT in OSD Manufacturing

Handling Challenging Powders in Tableting Operations

Pharma Industry Contends with Contamination-Based Recalls

Securing Each Dose: Reducing Falsification Risk with Dosage Level Authentication