	Targeted nanoparticle-based drug delivery has been a subject of interest for more than 20 years, and it offers a number of benefits over conventional treatment options. Encapsulation of an API in a carrier particle can protect it from degradation and allow its dispersion into an aqueous environment-the body-where typical APIs are poorly soluble. Furthermore, targeting groups can be used to address specific biological settings, maximizing the efficacy of the API while reducing the dose and, as a result, the potential for side effects.

	Important goals in nanoparticle production are ensuring homogeneous particle composition, minimizing particle size distribution, and maximizing API loading. Of these, particle size distribution has arguably the most significant implications for drug delivery, because nanoparticle size determines the rate of diffusion through a tissue, and different sized particles will be taken up by cells using different mechanisms. Release of the API-either by simple diffusion or nanoparticle degradation-will also be strongly influenced by size. Smaller nanoparticles will have a greater surface area-to-volume ratio and are therefore likely to release the drug much more rapidly. This rapid release may result in high API concentrations that could potentially lead to harmful side effects. Consequently, a broad size distribution means poor control over how the API is released, making it harder to determine whether or not the patient is receiving the required therapeutic dose. This lack of control is driving the demand for production methods that reduce polydispersity.

	Nanoprecipitation is the most common method for obtaining particles less than a micron in diameter. Industrially feasible production techniques have traditionally relied on a three-stage process: dissolution of a hydrophobic polymer in a waterâmiscible solvent, mixing of this organic phase with an aqueous solution, and precipitation of the polymer. Conducting the precipitation in the presence of surfactants-or using polymers that in themselves are surfactants-preventing polymer aggregation, and co-dissolving the API in the organic phase in the first instance leads to its encapsulation within the nanoparticles.

	Batch processes have typically offered the benefit of producing a large volume of material in a short period of time, as well as being conceptually easy to assemble. One-pot pouring or dropwise addition of the organic phase to the aqueous solution is the standard technique for nanoprecipitation, yet this simplicity is offset by a key disadvantage; it is difficult to set up or scale up a batch process with perfectly reproducible mixing. Even a trivial parameter, such as the distance between the magnetic stirrer and the point of injection of the organic phase, can have a profound effect on both the size dispersity and the average particle size.

In contrast to batch processes, microfluidics-based devices (see

), offer a higher level of control, because the mixing of liquids takes place in channels of controlled size and geometry, and almost invariably under laminar flow. In a cross-shaped microfluidic chip (Asia system, Syrris), for example, the organic phase passes through a central channel and concentrates in the middle region when water is added laterally 

 the two remaining perpendicular and counter-flowing channels. The mixing is relatively slow, laminar, and consistent, and this reproducibility makes the nanoprecipitation process easy to replicate. Furthermore, the size of the particles precipitated can strongly depend on the aqueous-to-organic ratio, which can be controlled in a microfluidic process. Finally, the production can be scaled up by running several microfluidic chips in parallel.

	The end result of this increased control is the reproducible production of homogeneous particles with a considerably narrower size distribution than most equivalent batch processes.  Homogeneity not only offers clear benefits for drug delivery, it also delivers significant upstream advantages. During research and development, it is easier to rationalize biological results when one is confident of consistent particle size, and it is easier to transfer the process to a good laboratory practice or good manufacturing practice environment, which is a necessary step for clinical translation of the product.

	The future of nanoprecipitation for drug delivery is likely to lie with flow techniques, but it is important not to downplay the role of batch processes, which are still key. The simplicity of batch techniques makes them ideal for exploring new materials or experimental conditions, and performing initial screenings without running the risk of, for example, obstructing the microfluidic channels. Once nanoprecipitation has been confirmed and refined under batch conditions, it can be transferred to a flow scheme for better reproducibility and control of the average nanoparticle diameter and size distribution.

	The continued development of the ability to finely tune nanoparticles for drug delivery will remain a key objective in this growing area of research. The microfluidics-based flow technologies now available to academia and industry offer an alternative to batch processes for reproducible generation of homogenous nanoparticles and will continue to be catalysts for innovation and experimentation in the future. 

olid Dosage Drug Development and Manufacturing

	When referring to this article, please cite it as N. Tirelli, "Improving Nanoprecipitation Control," 

	Nicola Tirelli is a professor and senior researcher at the Laboratory of Polymers and Biomaterials, Italian Institute of Technology.

Articles

Features

The performance of nanoparticles used as carriers in drug delivery is intimately linked to their physical properties. Nanoprecipitation is a common method for the preparation of drug-loaded polymer nanoparticles, but until recently, the reproducibility of the two primary dimensional descriptors-the average particle size and the breadth of the size distribution-has been a challenge due to the intrinsic variability of batch processes. Microfluidics-based flow techniques, however, reduce variation in drug-loaded polymer nanoparticle synthesis. 

Improving Nanoprecipitation Control

Reducing “time to market” is the ultimate goal for every pharmaceutical lab. Being the first on the market brings a competitive advantage for prescription, over-the-counter, or generic-drug manufacturers. Applying quality-by-design (QbD) principles at the formulation phase can prevent tablet defects at an early stage and thereby drastically reduce time during the complex and troublesome phase of “scale-up.”  Waiting until late in development in the “production-size phase” may force scientists to solve formulation issues at the pilot level or-even worse-in actual production. A QbD approach, however, secures the scale-up to production with maximum safety right from the beginning.

	Tableting instruments (i.e., compaction simulators) that simulate high-speed presses can be used in a quality-by-design (QbD) approach to perform in-depth material characterization and direct scale-up. Because they can modify the strain rate (i.e., the linear vertical punch velocity), tableting instruments can mimic the dynamics of a rotary tablet press running at full speed. The determination of the right material and quality attributes (e.g., lubrication, elasticity, cohesiveness, weight variation) can help in developing a robust formulation. An extensive characterization of a formulated blend can also prevent capping, sticking, or even die binding on a commercial-size rotary tablet press.

Lamination and capping are common tablet defects occurring in tablet manufacturing. Both terms are used to describe cracks on the side of the tablet. Lamination is a defect exhibiting cracks on the cylindrical part of the tablets (i.e., the “belly band”) as shown in 

	Capping is a defect occurring at the junction between the cylindrical part and the convex part of the tablet (see 

). Even though lamination and capping look more or less the same, some of their causes can be different.

 Lamination is due to air entrapment, as shown in work from the University of Bordeaux (1). An entrapped bubble of air begins to appear on the tablet surface at a pressure just below the pressure where lamination (cracks) can be observed. Applying pre-compression is then a efficient remedy. This de-aeration step will help remove the excess air. The ratio of pre-compression and main-compression can be studied. A pre-compression ratio of 10–30% is typically used in commercial-scale manufacturing.

	This air-entrapment can also come from a tight clearance of the compression tooling. Every manufacturer has its own mechanical tolerance between the punch tip and the die bore. However, a very tight tolerance is not recommended as the air will have a hard time escaping from the powder bed and will thus create air bubbles. Reversely, too large of a tolerance creates powder loss mainly on the lower punch.

	A tableting instrument can be used to troubleshoot or predict lamination issues. In a study performed with one of Medelpharm’s clients, a blend was compacted on a high-speed single punch tableting instrument using compression tooling from two different suppliers. Mimicking a Kikusui rotary tablet press at high speed, the tablets made with the first punch set had no lamination. The tablets made with the second punch set revealed lamination, although all the process parameters were identical with both punch sets. The cause of lamination was attributed to the difference of mechanical tolerances between the punch tip and the die bore. In this case, the tableting instrument was used to troubleshoot manufacturing issues and pin-point the parameter to be adjusted (i.e., change punch supplier).

	The effect of mechanical tolerances at the formulation phase is something formulation scientists could take into account. This example demonstrates that such process parameters should be considered in the first steps of QbD.

. Capping has its origin in the chemical nature of the excipients and APIs, the tablet shape, and process parameters, such as the turret speed, compression/edge thickness (and the resulting compression force), or insertion depth (i.e., penetration depth). Capping is ingloriously famous because it generally occurs during scale-up, either at the clinical manufacturing stage or during scale-up on a commercial-size rotary press. If tablet capping is discovered at a late stage, reformulation is most likely not an option anymore.

	The first process parameter that can be adjusted is the convexity of the tablet by modifying the radius of the punch tip to reduce capping tendency. Computer simulation (2) using finite element modeling has shown that a radial (i.e., shear) stress appears on the tablet cap when the upper punch tip is moving away from the tablet surface. The upper punch first loses contact at the land (i.e., the little flat portion surrounding the punch tip), which creates stress in the radial direction, explaining why capping occurs in the land region. Some experienced tableting experts know that the higher the curvature (i.e., the lower the radius), the higher the risk of capping tendency. Thus curvature becomes another process parameter to be evaluated in a QbD approach. 

	“Flattening” the tablet has its limits, however, especially when the tablets have to be film coated. Trying to coat flat-face tablets generally result in a defect known as “tablet twinning,” where two tablets are glued together.

	The second process parameter that can be adjusted after the tooling shape is the pre-compression. This will remove excessive air inside the powder bed and most likely enhance the cohesion of the tablets. This additional cohesion should most likely counterbalance the shear stress inherent to the tablet shape and avoid capping.

	The third process parameter is the insertion depth, also called upper punch penetration. By compacting deeper into the die, the applied pressure becomes symmetrical, thus densifying and creating cohesion equally on both sides of the tablet. Similar to adjusting pre-compression, the additional cohesion on the upper part of the tablet might be enough to prevent capping.

	A fourth process parameter is the compression/edge thickness (i.e., distance between the punch). By increasing the compression thickness, the compression force will be mechanically decreased and capping should disappear rapidly. The tablet breaking force (cohesion) will also drop, however, and it will most likely change the disintegration time and dissolution profiles as well. This process parameter needs to be assessed carefully.

	Adjusting all these parameters using a commercial-size press is time-consuming and requires large quantities of blend, but it is possible to evaluate capping by using single punch presses with high strain rate capabilities. The experiments described previously can be performed on such compaction simulators to troubleshoot tablet defects with small quantities of blend in a timely manner.

	Formulators can use QbD to optimize a formulation in the early development stage, before scaleup. Based on the quality target product profiles (QTPP) and the process flow chart (wet/dry granulation, tableting, coating), formulation scientists will have to list the material attributes (MA), quality attributes (QA), and process parameters (PP) that are required to achieve the QTTP. This risk assessment, based on the scientist’s process understanding and experience, shall then pinpoint the critical attributes and parameters and assess them with the compaction simulator.

	As described earlier on capping and lamination, the process parameters studied to troubleshoot the defects can be evaluated during formulation to determine the process space to produce good tablets without capping or lamination.

 Material attributes of the API and excipients generally include physico-chemical attributes, such as assay, impurities, particle size distribution, flow indexes, water content, and others. The compactibility of the ingredients, however, is not always taken into account for a simple reason: excipients have to comply with the monographs listed in the pharmacopeias, and these monographs do not contain any functionally-related specifications. Surprisingly, an excipient designed for direct compression does not have any specifications on its ability to form bonds, which is what should be expected from a binder. A scientist getting an 

United States Pharmacopeia (USP)/European Pharmacopoeia (Ph. Eur.) 

compendial excipient shall only rely on the supplier’s brochure on its performance in tableting. This is the same for an API for which it could be possible to test its ability to form bonds under pressure.

	A generic-drug manufacturer, for example, that intends to source an API from different drug substances suppliers should consider various properties. In addition to the chemical purity criteria and other common physical characteristics, such as particle size distribution or specific surface area, it is wise to make a tabletability profile on an instrumented tablet press. Due to the poor flowability of APIs and small quantity of available API at this stage, the loading of the die would most likely be carried out manually. (Note that external lubrication with a dry lubricant on the die bore and punches is often necessary to avoid sticking and die binding.) If the API is able to form bonds, it’s then possible to plot the tensile strength vs. axial pressure, as defined by 

 Chapter <1062>, which was introduced in June 2017 (3). This tabletability profile can be used to compare the different grades of API, and can help choosing the right grade for the drug product. This approach can be performed the same way on neat excipients.

 Evaluation of lubrication and the determination of criticality of certain material attributes should also be performed. It is widely thought that a quantity of 0.5–1% of lubricant is necessary in the tablet formulation. But is this correct? The obvious quality attribute to look at is the ejection force. However, there are other QAs that can be studied. First, the ejection force is only the peak of the complete ejection force signal. By taking a close look at the signal, it is possible to see oscillations on the signal just after the peak (see 

). Even if the peak of the ejection force is still fairly low, this is a sign that die binding (also known as die tightness) is occurring. A less common approach is to consider also the transmission coefficient (4), defined as the ratio of the upper and lower punch force. To measure those forces, an R&D press will have to be equipped with force sensors on both punches and be able to operate the punch in a non-symmetrical way.

	Older, common technologies, such as eccentric R&D presses, can do the trick if they are well instrumented. The compression force recorded by the lower punch will be systematically lower than the force recorded by the upper punch. The powder densification occurs first at the upper side of the powder bed. The energy provided to the system will be partially lost due to friction between particles and between particles and the die bore. This energy loss will result in a measurement of a lower punch force. The target of the transmission coefficient should be between 90% and 100%. A low transmission ratio, such as 70%, might be linked to ineffective lubrication. By looking at the peak of the ejection signal, the oscillations of the ejection signal, and the transmission ratio, the quantity of lubricant and its associated blending process can now be optimized. Different grades of magnesium stearate, a well known lubricant, featuring different specific surface areas, can give very different lubrication.

 Elastic recovery is another parameter seldom assessed. Acquiring these data requires the tablet press to be instrumented with position sensors. The elastic recovery is the difference between the tablet thickness measured out-of-die, with a caliper for instance, and the in-die tablet thickness measured by the sensors at the peak of compression. Elastic recovery is often linked to lamination as it can create micro-fractures within the tablets. Interparticular cohesion is therefore reduced and lamination can occur. As an example, calcium phosphate excipient exhibits an elastic recovery around 4%. But some sustained release polymers can be as high as 20%. Generally speaking, it is recommended to associate ingredients having similar mechanical properties, especially when formulating bi-layer tablets where an elastic layer could induce a layer separation.

 The compression force is quite often considered as a process parameter. Actually, it is first a quality attribute. On a basic rotary tablet press, an operator can adjust the dosage height (and its corresponding quality attribute “tablet weight”) and the compression/edge thickness. The compression force is then measured by strain gauges located on the pressure rolls. Decreasing the compression thickness will result in increasing the compression force and vice versa. That is the main reason why many people think that this compression thickness knob is controlling the compression force. Now, when the operator increases the dosage height, the compression force will also increase. In this case, compression force cannot be a process parameter and is in fact a quality attribute.

	On the other hand, modern rotary tablet presses are equipped with a “weight control loop”. This control loop will basically rely on the relation that exists between the tablet weight and the compression force. (One exception is GEA, formerly Courtoy, which uses the relation between tablet weight and tablet thickness.) The strain gauges measuring the compression force are the indicators to monitor the tablet weight. Any variation of the compression force will be an indication of a variation of tablet weight, most likely due to a non-uniform blend density and flowability between the beginning and end of the batch. A control loop will then electronically change the dosage height to maintain the compression force within the  target value (i.e., set point).  A production press is mechanically designed to compress the powder bed to a given volume, ensuring that similar force indicates similar weight.  In this case, the particular set point for compression force is a process parameter. Depending on the context, compression force is both a QA and a PP.

	Considering compression force as a QA can help a formulator speed up tablet development, by plotting the relation between the compression force and the tablet weight. To do that, the PP “dosage height” has to be modified to mimic a change in powder density during the process. For example, if the nominal tablet weight is 850 mg, the dosage height can be adjusted to reach 850 mg + 5% and 850 mg – 5%. Tablet weights within this range are compliant with the uniformity of mass test as set forth by the 

(5). The scientist can now plot the compression force versus tablet weights (see 

). This graph will be crucial to help set up the ejection and tolerance set points on the commercial-size rotary tablet press during scale-up, thus saving time and material. In addition, other QAs, such a tablet breaking force (also known as “hardness”), disintegration time, or even some key dissolution times can be plotted versus tablet weight.  All these graphs will guide the formulator in the determination of the design space.

	This full QbD approach has been implemented for complex oral solid dosage forms, such a multi-layer tablets or tablet-in-tablet, at several contract development and manufacturing organizations. Using a tableting instrument with high speed rotary press mimicking features, the so-called compaction simulator, allows design of robust formulations, smooth scale-up, and reduced risks and costs, ultimately accelerating the time to market.

 478 (2) 702–4 (2015).
	2. V. Mazel et al., 

USP General Chapter <1062> Tablet Compression Characterization 

(US Pharmacopeial Convention, Rockville, MD, 2017).
	4. T. Ménard and L. Pisarik, 

EurPh, General Chapter 2.9.5, Uniformity of Mass of Single-Dose Preparations 

	When referring to this article, please cite it as R. Cazes, "A QbD Approach to Shorten Tablet Development Time," 

Solid Dosage Drug Development and Manufacturing

	Regis Cazes is strategic marketing director at Medelpharm, France, a company creating the STYL’One family of tableting instruments and providing contract lab services in powder characterization and tableting; 

Tableting instruments (i.e., compaction simulators) that simulate high-speed presses can be used in a quality-by-design (QbD) approach to perform in-depth material characterization and direct scale-up.

A QbD Approach to Shorten Tablet Development Time

During early drug development, compounds have to be characterized for their toxicology, bioavailability, pharmacokinetics (PK), and pharmacology profiles. To carry out characterization studies, the compound must be dissolved in solution at a concentration that is high enough to achieve a therapeutic effect. Low aqueous solubility can, therefore, be a major challenge in drug development. 

	Methods to improve solubility include chemical modifications, physical modifications, and solvent modifications. The method selected will be based on the compound’s chemical properties, the physical state of the formulation, and the route of administration. In the development of liquid formulations, solvent modifications and carrier systems are most commonly used because they affect only the solvation characteristics of a drug rather than its solid-state properties. 

	Excipients can be used to optimize the solubility of poorly soluble compounds. Traditionally, the selection of excipients relies on a trial-and-error-based approach involving a number of research-based methodologies to determine the ideal excipients. Although often successful, in reality, this approach is time-consuming, costly, and demands large amounts of material.

	Developing a high-throughput screening method

	A high-throughput screening method has been established to overcome the challenges of selecting one or more excipients. The aim was to develop a methodology that would use minimal amounts of API, while providing a cost-effective and efficient way to achieve results. It was also a prerequisite to establish a platform that could provide conclusive information about a compound’s chemical stability in varying solvents and excipients. 

	Several experiments were performed to establish the methodology. The screening list involved excipients with varying solubilization mechanisms, including water-soluble organic solvents, non-ionic surfactants, water-insoluble lipids, organic liquids/semi-solids, cyclodextrins, and phospholipids.

	The type of excipients used will depend on the drug delivery system. Orally administered compounds, for example, will require different excipients than injectables. There is also a need to ensure that the final concentration of the selected excipients is within the generally recognized as safe (GRAS) list of recommended concentrations. The identification of the correct excipient in its individual correct maximum concentration is important, especially for parenteral formulations, because doses that are too high can cause pain, hemolysis, or inflammation.

	The high-throughput screening platform is based on identifying the solubilization capacity of each excipient for a compound. It can also shorten the time taken to identify an excipient by allowing multiple tests to be performed simultaneously.

	The method was initially developed using six commercially available drugs (see 

) with diverse chemical properties. Testing was conducted using 30 excipients dispensed in 96 well-plates via a fully automated robotic system (Tecan). Three plates were studied for each compound. The plate was shaken for 48 hours to achieve equilibrium. The results were compared with solubility measurements performed using a manual shake flask method where 15 mg of powder and 2 mL of excipient were added. The samples were again shaken for 48 hours, centrifuged, and then analyzed by high-performance liquid chromatography (HPLC) to determine solubility and detect any degradation. The measurements were performed in triplicates.

	Some excipients have been shown to offer better solubilization capacity than others; the trend varies between compounds (see 

). For ionizable compounds, pH-dependent solubility is a useful approach, especially if it can be combined with another solubilizing excipient. The contribution of solid-state barrier to solubilizing a compound appears to be more pronounced at a cut-off level of solid-state properties. Before this cut-off, the solubilization of the compound was more compound specific, which creates the need to also test on a larger set of excipients. 

	The results of the high-throughput screening method demonstrate that solubility using this technique is not statistically different than that achieved when using a manual approach. The method can provide information on the solubilization capacity of compounds in different excipients, while also offering insight into stability.

	The high-throughput screening method overcomes the challenges associated with manual approaches by being more cost-effective and economical in the use of materials, while turning around results in three to five days per set of compounds.

	The development of the platform has created new possibilities for reduced drug development timelines and costs. Furthermore, the information generated in the screening is useful in the later stages of formulation development. Helping to boost the probability of successful formulation, selecting excipients based on an API’s unique molecular properties offers a faster process that can revolutionize the way developers evaluate the solubility of any compound.

	When referring to this article, please cite it as A. Alhalaweh, “High Throughput Screening of Excipients," 

Pharmaceutical Technology Solid Dosage Drug Development and Manufacturing

	Amjad Alhalaweh, PhD, is a formulation scientist at Recipharm.

The author discusses the advantages of high-throughput screening of drug formulation excipients by a fully automated Tecan-robotic system.

High Throughput Screening of Excipients

The development of effectively formulated oral drug products is key to ensuring patient compliance and desired clinical outcomes. Given the increasing regulatory expectations for the development of bespoke age-appropriate products, critical performance features now include palatability and overall acceptability in addition to standard drug-delivery requirements for oral medications. Understanding and modifying the taste attributes of aversive drug substances is, therefore, important. There are several taste-masking techniques available, which involve either modification of the API itself or the formulation; however, there are as yet no standardized industry approaches for assessing whether the poor taste of a drug has been effectively masked.

 and preclinical methods can be applied, the resulting data are, at best, incomplete and, at worst, misleading. Approaches that have greater correlation with human response and, ideally, incorporate clinical assessment of the formulation are, therefore, required. This article reviews the breadth of taste-masking techniques available, the methods used to evaluate taste, and how an integrated approach to formulation development and clinical assessment can deliver significant benefits for product development and validation. 

	Compliance and the need for taste masking

	The growing industry interest in palatability and acceptability of medicines is driven primarily by issues around patient adherence and compliance. Compliance is a particular issue with pediatric and geriatric patients (1), with certain medicines only achieving 11% compliance in children (2). These populations are not only the most sensitive to taste, but are also the patient groups who suffer most from dysphagia or have difficulty in coordinating swallowing, making the need to generate age-appropriate medicines imperative.

	Regulators are now stipulating the requirement for pediatric investigation plans (PIPs) and pediatric study plans (PSPs) for all new registered products, making compliance essential to development. Demand is also growing at unprecedented rates amongst the ageing population. The World Health Organization estimates that the number of people aged 60 years or over is predicted to grow to 1.4 billion by 2030, and nearly 2.1 billion by 2050--more than double the number in 2015 (3). The so-called “silver tsunami” is poised to become one of the most meaningful social transformations of the 21st century. 

	The palatability of a medicine is largely dictated by the taste of the API. A significant percentage of APIs on the market or in development are bitter tasting or unpalatable. In most cases, taste issues can be overcome by simply formulating the API into a tablet or capsule. This approach, however, cannot be used as standard in pediatric and geriatric populations because of swallowing difficulties associated with oral administration. Here, alternative formats such as liquids, suspensions, or dispersible or chewable products must be used.

	A number of physiological and physicochemical approaches have been employed in an attempt to mask the taste of APIs and/or to prevent drugs from interacting with taste buds. These approaches fall into four main categories, outlined below and discussed in more detail in the paper published by J. Walsh 

. API modification involves generating a new solid form or salt of the API, or administrating it as a prodrug. These approaches, however, are not always viable as the API may have just one stable form, or it may not be possible to form a salt with the neutral API.

. Components such as sweeteners, flavorings, or solubility modifiers can be added to the formulation to overcome the taste of the API. This method is generally problematic for high dose APIs, where it may not be possible to mask the taste.

. The API can be complexed with a number of ligands to prevent the molecule interacting with taste receptors. These ligands include cyclodextrins, ion exchange resins, and polymers. This method is only possible with relatively low drug loading. The potential risk of altering the pharmacokinetic (PK) performance should also be considered.

. Coatings can be used to effectively cover the drug product and prevent the API from being released in the mouth. Coated formats, however, do not overcome issues with swallowing and lack of compliance. In addition, coatings add significant cost to the production process and may also affect the release and PK profile of the drug.

	Beyond palatability and API properties, it is also necessary to consider broader drug-delivery needs when designing a taste-masked formulation. Factors such as requirement for solubilization enhancement; excipient stability limitations; patient age (and therefore the acceptable daily intake of excipients); whether the API must be taken with water or food; and impact on storage requirements (i.e., the need for refrigeration), all make the design of taste-masked formulations more complex, and the effective assessment of taste more important.

	There is no standard industry approach defined for assessing and verifying whether a drug product’s taste is acceptable. Although there is a range of commonly used techniques, outlined in 

, the lack of standardization presents a significant issue in determining the taste of drug products.

	All of the techniques, apart from human taste panels, are problematic in that they are surrogate methods, and, therefore, do not necessarily predict or match the human response. There is also additional time and cost associated with formulating and performing these surrogate analytical or preclinical tests. If the formulation is subsequently deemed unsuitable from a taste perspective in humans, further development cycles may be required, adding yet more cost and time to the process. It is, therefore, imperative that formulation selection is based on clinical taste assessments. Limitations can still be apparent if an acceptable formulation is not identified from the initial raft of prototypes prepared for human testing, highlighting a preferred need for a test model that allows flexibility to make compositional adjustments in real-time based on arising sensory data.

	Rapid development and assessment of taste-masked products

	The integration of formulation development, real-time adaptive GMP manufacturing, and clinical testing has been successfully used for both the assessment of taste and sensory attributes of drug substances, as well as the rapid screening of prototype taste-masked formulations. This model reduces development time and cost (given drug products are prepared within hours or days of dosing), and maximizes the potential for success, given adjustments to compositions are based on arising human clinical data (e.g., safety, tolerability, PK, pharmacodynamics [PD], or taste). As shown in 

, flexibility can be enhanced by up-front definition of a formulation design space with bracketing ranges in the levels of critical-to-performance excipients.

	To perform the taste assessments, a consumer preference panel is established, typically of 12-18 subjects, using healthy adult volunteers trained in “sip and spit” tasting techniques. The emphasis of the panel is on “preference” rather than “measuring levels” of taste, providing representative data on acceptability of a formulation to a general population. Participants complete a bespoke questionnaire, consisting of visual analogue or hedonic scales to characterize a variety of API and formulation parameters, an example of which is shown in 

. The scale will typically give seven ranking levels, ranging from “strongly like” to “strongly dislike.” Several formulations can be assessed within a single day, with a one- to two-hour gap between tasting events.
	

	The program design and taste-assessment protocol are customized to the specific API and formulation in question. Design space variables can be established for API properties (e.g., particle size) or formulation attributes (e.g., levels of sweeteners, flavors, or viscosity modifiers). The protocol can also include assessments of API only (at single or multiple concentrations), positive controls (e.g., quinine for bitterness), the inclusion of replicate assessments to serve as further controls, and/or the use of multiple study periods to allow for interim analysis of data to guide decision making.

	Performing these studies within the United Kingdom requires submission to and approval from an Independent Ethics Committee (IEC). Whether a regulatory review by the Medicines and Healthcare products Regulatory Agency (MHRA) is required will depend on study objectives, specifically if safety assessments are required.

	Combining taste and pharmacokinetic evaluation

	While taste assessments can be used as the sole clinical endpoint, they can also be combined with PK measurements as part of the same study. This approach can be particularly important if the taste-masking strategy has the potential to affect the PK performance of the existing (adult) formulation. The drug half-life requires a washout period, meaning that a taste evaluation can be incorporated into the study design without delaying the product assessment cycle (which is typically one to two weeks). The result is a powerful combination of clinical data confirming product palatability and acceptability, coupled with a full understanding of the PK performance in humans. In combination, this presents an ability to transition to efficacy studies in the target patient population with confidence, with an appropriate formulation, and with an informed dosage regimen.

	The following case study demonstrates how the Quotient Sciences approach facilitated the rapid redevelopment of a taste-masked formulation for the long-term treatment of hyperkalemia (6).

	Development and assessment of taste-masked formats for chronic disease

 Patients with heart failure or chronic kidney disease are at high risk of developing the potentially life-threatening condition hyperkalemia. The current approved treatment, sodium polystyrene sulfonate, has poor palatability and is unsuitable for long-term use. RDX7675 is a novel product being developed for the treatment of hyperkalemia. RDX7675 is a structural derivative of sodium polystyrene sulfonate and, therefore, shares some issues with taste and palatability. The goal was to expedite development of a clinically validated formulation for RDX7675, which suitably masked its taste, to enable its long-term use in the treatment of hyperkalemia (6).

 Rapid screening of multiple formulation types and flavors was undertaken. All formats were manufactured within 24 hours of dosing requirements. A flexible clinical protocol ensured that any required changes to the formulation, flavor, or viscosity could be made within the study. A total of 18 subjects were included on the consumer preference panel in the two-period clinical study. Formulations were tasted every two hours, with doses expectorated, and palettes cleansed with water and unsalted crackers between tastings. Data were captured via questionnaires.

. The flexible study design enabled real-time manufacturing and rapid taste assessments of multiple formulation options. Data from the consumer preference panel were available within nine weeks of program initiation. These data enabled the selection of a lead formulation to progress to the pivotal PK/PD clinical study (

	The need for taste-masked products continues to grow as a result of the significant number of unpalatable drugs in development and the requirement for age-appropriate delivery formats. The in-vitro and preclinical methods used to assess whether a formulation has effectively masked the taste of an API show poor correlation with human response, particularly where pediatric medicines are concerned (7). Where sensory human taste panels are used to quantify specific criteria with great precision, highly trained subjects are required, with the associated additional costs and timelines.

	By integrating formulation development, real-time adaptive GMP manufacturing, and clinical assessments using consumer preference panels, formulations can be rapidly screened and their taste characteristics assessed. Acceptable taste-masked formats can be identified and optimized in real-time based upon arising sensory and/or PK data. Generated data can be used both to inform further development, as well as support compilation of PIPs and PSPs.

Periodic Survey #44 Patient compliance with prescription regimens

,” 2015.
	4. J. Walsh et al., Ad. Drug Deliv. Rev. 73 14-33 (2014).
	5. M. S. Bahia 

. published online Mar. 28, 2017, doi.org/10.1016/j.ijpharm.2017.03.076.
	6. V. Zann 

. 6 (11) 2663-2673 (2017).
	7. J. A. Mennella 

	When referring to this article, please cite it as P. Scholes, “Removing the Bitter Taste from Drug Development," 

	Peter Scholes is chief scientific officer at Quotient Sciences.

Integration of formulation development, real-time adaptive GMP manufacturing, and clinical testing using a consumer preference panel can save time and cost in performing taste-masking assessments.

Removing the Bitter Taste from Drug Development

Pharmaceutical businesses are rapidly becoming global, with approximately half of the industry’s growth in emerging markets (1). A pharmaceutical company’s objective is, therefore, to build a sustainable and cost-efficient supply chain that meets global regulatory requirements. The selection of drug substance regulatory starting materials (RSMs) and justification of their designation in the pharmaceutical supply chain has become an industry-wide focus as a measure of mitigating regulatory risks and preventing unexpected rises in cost when transitioning from clinical to commercial supply chain. This article discusses the expectations of regulatory agencies, the scope of the starting materials’ presentation in the filings, and the risk and readiness for a “push back” (i.e., extension of the GMP portion of the API process upstream) in the event of an agency’s major objection to the sponsor’s RSM designation.

	The pharmaceutical supply chain can be divided into four segments, as shown in 

, from starting materials (i.e., the RSM), to the drug substance, the formulated drug, and ultimately, the packaged and labeled finished product. 

	Commonly, the emphasis during pharmaceutical development is on the portion downstream of the RSMs. The part of the supply chain upstream of the starting materials does not receive the same level of attention--less time is allocated for development of process and specifications, which can, therefore, present a regulatory risk and potential for an unexpected rise in cost when transitioning from clinical to commercial supply chain. One particular risk factor is the development of RSMs, or the lack thereof.

	Typically, at the time of marketing application submission, the sponsor would have validated the production processes of the drug substance, formulated drug, and finished product. This validation involves successfully conducting at least three consecutive commercial-scale batches for each segment of the supply chain and releasing each batch against pre-defined acceptance criteria (i.e., commercial specifications). If, at this advanced stage of development, the proposed RSM designation is not accepted by a regulatory agency, then the entire supply chain that stems from this RSM is essentially invalidated. The sponsor will likely experience lengthy delays to address RSM designation issues before the marketing application is approved.

	One of the most underrated risks is for sponsors to assume that RSM designations used throughout clinical trial applications will be accepted in the marketing application without justification. An industry-wide disagreement on terminology--resulting in a plethora of names related to starting materials, such as raw materials, key raws, critical raw materials, key raw materials, critical raws--further confuses the role of starting materials in the supply chain. To clarify, an API RSM could be a raw material, a manufacturing process intermediate, or even an API. The material could be an article of commerce that is available from multiple sources, typically in large quantities and often sourced under commercial agreement. The material could be produced in-house and manufactured using custom-designed process. The RSM should be used in the production of API and it should represent a significant structural fragment of the API’s chemical structure. The sponsor of the clinical or marketing application should designate the RSMs and document rationale for their selection (2). The RSM’s designation marks the point at which GMP production (described in Section 3.2.S.2.2 of CMC module 3) (3) of the API begins, as shown in 

	In 2014, the European Medicines Agency (EMA) published a reflection paper on the requirements for selection and justification of starting materials for the manufacture of chemical active substances (4). EMA felt that the current guidelines lacked detailed specifics, thus leading to a variety of interpretations. Proposed starting materials specifications were often insufficient. EMA even encountered instances where starting materials were not discussed in the application or processes by which the starting materials prepared were not part of the overall criticality appraisal. More recently, the International Council for Harmonization (ICH) Q11 guidelines implementation group published two documents of questions and answers to clarify the ambiguity and provide additional examples for the selection and justification of starting materials (5). These documents are extensions of the original Q11 guidelines (6).

	As a guidance to industry, EMA included seven examples of critical manufacturing steps that should be performed under GMP:

		Steps involving the formation and/or purge of key impurities 

		Steps that introduce key structural features of the active substance 

		Steps requiring careful control of process parameters 

		Steps that use or generate genotoxic compounds 

		Steps that involve the use of class 1 solvents and/or toxic metals

	While the last example in this list pertains to the final isolation and purification step of drug substance, which should be performed under GMP setting, the first six examples are quite broad and could apply to the non-GMP portion of the drug substance process where RSMs are produced. For instance, “careful control of parameters” is expected in any manufacturing process. Furthermore, the definition of “complex chemical transformation” is somewhat subjective because any chemical reaction will have its own intricacies. Thus, the probability of EMA’s push back of a proposed RSM designation is quite high.
	

	Reasons for rejection of RSM designations

 shows the API portion of the supply chain and frequently encountered reasons for regulatory agencies to reject proposed RSM designations. It is important to note that the whole supply chain is impacted.

	The most common reason for rejection of a RSM designation is the insufficient number of steps (#1 in 

) in the API manufacturing process (CMC Section 3.2.S.2.2)  (3). Lack of in-process controls or inadequate acceptance criteria for formally released GMP intermediates (CMC Section 3.2.S.2.4) (3) represent inadequate quality control and is another common reason (#2 in 

) for the rejection of a RSM designation.

	The insufficient presentation of synthesis and controls for RSMs (CMC Section 3.2.S.2.3) (3) could also be the reason (#3 in 

) for rejection of a RSM designation. Regulatory agencies expect API-like acceptance criteria for RSMs, thus insufficient scope of specifications is another reason (#4 in 

) for major objection to a RSM designation. Lastly, the reviewer may determine that there was insufficient appraisal of criticality in the full synthetic route (i.e., from raw material or building block to API), leading to rejection of the RSM designation (#5 in 

	Successful defense of RSM designation is not guaranteed; therefore, the sponsor should preemptively work on three areas to prepare for possible push back: chemical synthesis, analytical controls, and manufacturing.

	Chemical synthesis plays an important role because the extent of push back will depend on the synthetic route. As part of the synthetic route scouting strategy early on in process development, the sponsor should identify an earlier intermediate as a back-up RSM. The push back will result in an increased number of GMP steps upstream, requiring in-process analytical controls and adequate scope of acceptance criteria for the back-up RSM.

	The rejection of RSM designation can lead to two scenarios. If the RSM was produced at a non-GMP plant, the sponsor would have to transfer production to a GMP plant, which can be stressful, especially if the sponsor learns about the major objection to the RSM designation in a pre-new drug application meeting or during marketing application review. The sponsor would be in a more favorable situation if the rejected RSM was produced at a GMP plant. They would then have an opportunity to retroactively validate RSM process at the same manufacturing site, in the same equipment, and on the same scale, and retain all commercial inventory, including all drug substance and formulated drug batches already derived from this RSM.

	Case study: addressing FDA’s push back and retaining launch inventory

	The following case study is presented to illustrate use of a back-up RSM. A schematic presentation of validated commercial process for the preparation of API is shown in 

	The proposed RSMs 1 and 2 were produced using a custom designed manufacturing process (i.e., neither of the RSMs was an article of commerce). RSMs 1 and 2 constituted 38% and 31% of the API’s core atoms, respectively. The core atoms are defined as all API structure atoms in the required connectivity and spatial orientation, excluding hydrogen atoms. Hence, RSMs 1 and 2 could be viewed as custom designed building blocks of similar complexity. Yet because RSM 1 was an additional step “away” from the API formation, the two RSMs had different propinquity to the API (see 

). As a result, the API GMP process lacked symmetry in its synthetic hierarchy.

	FDA accepted the proposed RSM 1 designation, but rejected the proposed RSM 2 designation, stating that the latter is used in the API making step, and thus is considered an advanced process intermediate and should be produced under GMP. The sponsor had anticipated a push back and had identified a precursor to RSM 2 as a potential starting material. As a risk mitigation measure, the sponsor developed appropriate GMP-level analytical controls and acceptance criteria as back-up preparation for RSM 2. Furthermore, the sponsor had produced all RSM 2 batches in a GMP plant, creating an opportunity to retroactively validate RSM 2 process in the same equipment, on the same scale, and at the same manufacturing site, hence, enabling retention of all downstream commercial launch inventory originating from RSM 2 to the API and formulated drug to the finished packaged and labeled product.

	In the updated API process, RSMs 1 and 2 constituted 38% and 20% of API’s core atoms, respectively, and could be viewed to have similar complexity, custom-designed building blocks with identical propinquity to the API (see 

). As a result, the API GMP process had symmetrical synthetic hierarchy.

	In another scenario, there is a possibility that FDA could accept the proposed RSM designation, while EMA rejects that same designation. The sponsor could accept the major objection to the RSM designation and extend the GMP portion of the API process upstream. In this scenario, the sponsor would be forced into a complex, two-stream supply chain, one for the US market and another for the EU market. Alternatively, the sponsor could overcome objection by presenting a comprehensive assessment of the RSM’s impact on the CQAs of either API or formulated drug.

	CQAs are justified by an impurities assessment. There are two types of impurities: 

		Those structurally related to the API, such as starting materials, process intermediates, and their by-products.

		Those structurally unrelated to the API, such as reagents, solvents, and their by-products.

	A laboratory-based R&D study, where high levels of process materials and impurities are purposefully introduced in the process (i.e., “spiking”), can assess an impurity’s fate by measuring their residual levels after one or more operations. Data from such studies can be used to calculate the estimated maximum level of any material in the API. Laboratory results are typically confirmed by analyzing manufactured batches of the API and corresponding process intermediates.

	The fate analysis of a RSM in the process and assessment of its impact to the API’s CQAs is presented in the example in 

	The API manufacturing process consisted of four distinct manufacturing steps. A high quantity of RSM was spiked in each step and the amount that remained upon completion of the step was measured. Individual purging factors (see 

) were calculated by dividing spiked quantities with the amount that remained. Iterative multiplication of all individual factors yielded a cumulative purging factor of 2.32 x 10

. To help reviewers appreciate how effectively the RSM is purged in the process, the maximum estimated level for this material in the API was calculated. The actual kilogram quantities of the RSM (100 kg) input and corresponding API (61.32 kg) output were used. A conservative assumption was made that only approximately 80% of the RSM will convert to the API, leaving approximately 20% of unreacted material, corresponding to 20.14 kg. Division of this residual amount by the cumulative purging factor (2.32 x 10

) yielded the maximum estimated level, 8.68 x 10

 mg, in the API. This quantity was converted to a more common presentation of concentration for an ultra-low-level impurity: parts per million (ppm = mg/kg). Division of the maximum estimated level (8.68 x 10

 mg) by the API batch output (61.32 kg) yielded 1.42 x 10

	The purging study findings were confirmed by analysis of eight API batches with a method that could detect residual RSM at a level as low as 0.05 ppm. In summary, the results from analysis of laboratory-scale purging studies and manufactured batches conclusively demonstrated that the proposed RSM was purged in the process and had no impact on the API’s CQAs. The study results enabled the sponsor to exclude the analysis for this RSM in the API acceptance criteria. Agencies worldwide accepted the justification for this RSM.

	RSM designation justification: clinical trial versus marketing application

	The lesson learned is that regulatory agencies tend not to challenge RSM designations in clinical trial applications because they do not want to impede clinical development. Thus, unless the GMP portion of the API process is very short (for instance, where there are no chemical bond-making reactions, but just purification steps), the agency is not likely to reject the proposed RSM designation in the clinical trial application. Expectations are, however, very different as sponsors transition from clinical to commercial supply chain. The sponsor should, therefore, be prepared to provide a rationale and evidence in support of the RSM designation.

	In conclusion, the CMC sections in regulatory filings should first demonstrate that the sponsor understands the science and technology used to produce the drug and can provide quality data to support statements in the application. Second, the sponsor has to show that adequate controls are in place throughout the entire process and that at no point in the supply chain are patients put at risk.

Rethinking the Pharma Supply Chain: New Models for a New Era

,” Boston Consulting Group (May 2013).
	2. FDA, 

Guidance for Industry, Q7 Good Manufacturing Practice Guidance for Active Pharmaceutical Ingredients

, (Silver Spring, MD, September 2016).
	3. FDA, 

Guidance for Industry: M4Q, the CTD, Quality, U.S. Department of Health and Human Services Food and Drug Administration

 (Center for Drug Evaluation and Research, 2001).
	4. EMA, 

Reflection Paper on the Requirements for Selection and Justification of Starting Materials for the Manufacture of Chemical Active Substances

 (London, September 2014).
	5. ICH, Q11 Implementation Working Group “

Development and Manufacture of Drug Substances (Chemical Entities and Biotechnological/Biological Entities), Questions and Answers (regarding the selection and justification of starting materials)

Guidance for Industry, Q11 Development and Manufacturing of Drug Substances

 (Silver Spring, MD, November 2012).
	7. M. Faul et al., 

, 18 (5) 594-600 (2014).  
	8. M. Faul et al., 

	When referring to this article, please cite it as V. Jurkauskas and M. Chen, “Transition From Clinical to Commercial Supply Chain-Regulatory Starting Materials," 

, is vice-president and head of Chemistry, Manufacturing and Controls (CMC) at Akebia Therapeutics and 

, is CEO of STA Pharmaceutical, a WuXi AppTec company.

The authors discuss expectations of regulators on the selection of drug substance regulatory starting materials (RSM) and the justification of their designation in the pharmaceutical supply chain, the scope of the RSMs' presentation required in regulatory filings, and how to mitigate and prepare for "push backs" in the event of a major objection to the sponsor's RSM designation.

Transition From Clinical to Commercial Supply Chain-Regulatory Starting Materials

Based in Dublin, Innopharma Labs was launched in 2009 during the heady days of FDA’s Process Analytical Technology (PAT) initiative, when advocates of modern process control and industrial engineering predicted a major shift in the industry’s approach to quality control and manufacturing. 

	Years later, a full-scale shift may not have yet occurred, but concepts such as continuous manufacturing and pharmaceutical quality by design (QbD), and a redefinition of process validation, have begun to drive new, more modern approaches to drug development and manufacturing with an increasing number of approvals implementing these principles.  Moving beyond its foundation in PAT and particle size imaging, Innopharma has responded by branching out into new areas that include automated process control, services, and education.

	In January 2018, the company formally launched a new advanced manufacturing program based on the concept of predictive control, permitting “self-guided” granulation and sophisticated control of modified-release oral solid dosage forms. Enabling this move was a  €13-million investment in Smart FB, a suite of technologies that include a data historian and cloud-based data repository, to allow manufacturers to achieve the benefits of more advanced process control for batch processes.  The technology can be test driven at the company’s new dedicated process R&D facility in Dublin.

	Company founder and CEO Ian Jones discussed these initiatives and industry trends with 

	While Innopharma is focused on the pharmaceutical industry, particularly formulation and manufacturing, it also serves the chemical, food processing, and solids processing sectors, and this knowledge informs the company’s technology platform. Jones will be discussing Manufacturing 4.0 at INTERPHEX 2018 in April.

	Exploring self-guided granulation and coating

: When did you launch your advanced manufacturing pilot program and are you currently working with any industry partners?

: We officially launched the pilot program in January 2018, after testing it with pharmaceutical companies during 2017.  This work builds on the application development and sensor system integration with partners Glatt, a specialist in fluidized bed drying, and Colorcon, which focuses on formulation materials including coating.

: Are you working with any academic partners?

: Over the past seven years, we have collaborated broadly with academia for PAT, but decided to invest in hiring process scientists, engineers, and chemometricians and bring this expertise inhouse. Based on input from pharmaceutical manufacturers, we have also developed our own Smart FBX advanced development and manufacturing modular platform incorporating PAT sensors, data integration engine, secure data storage, analytics and automated control components.

: Why did you feel this was necessary?  After all, can’t companies currently source the individual technologies (e.g., data historians and process control platforms) separately, themselves?

:  We felt that it was important to bring everything together. As mergers and acquisitions have become the norm in pharma, companies are managing a more diverse and global network of manufacturing sites than ever before, and contract manufacturing organizations (CMOs) are becoming a more prominent part of these networks. Our goal is to provide process data visibility, and to enable global access to development and manufacturing data so that we can support process development projects, wherever in the world they are based. Where companies lack in-house resources, our staff can step in and help with development and control problems.

: Why do you see advanced control as being so important to pharma’s future?

: Just consider a car or cell phone from 15 years ago and compare them to what exists today. They are almost completely unrecognizable. A high level of advanced analytics and diagnostics has been built into these devices which has become standard in these industries, but this is not yet the case for pharma.

	If the pharmaceutical industry wants to develop and manufacture safer and more affordable medicines faster, improved automation and process control will be key enablers.  That’s why we’ve gone down this road. At this point, the goal is not to pioneer, but merely to start to catch up to what other industries are doing with automation and analytics.

	Smarter control will be especially crucial for scaleup and tech transfer.  More, and better, technology is needed to support this work in the pharma industry.

:  What do you see as being responsible for pharma’s conservatism?

:  The main reasons are financial and regulatory. Today, the basic focus is on doing more with what you have, rather than investing in new technology. There is also an understandable conservatism to moving away from the traditionally accepted regulatory path in case of delay to approval or a request for significant additional data for a variation submission. As a result, it has been a major challenge for tech providers and the industry to get to a stage where advanced manufacturing is even possible.

Can an Irish analytics company and its CEO bring pharma closer to 21st-century practice?

Pharmaceutical Technology-03-01-2018

Improving Nanoprecipitation Control

A QbD Approach to Shorten Tablet Development Time

High Throughput Screening of Excipients

Removing the Bitter Taste from Drug Development

Transition From Clinical to Commercial Supply Chain-Regulatory Starting Materials

Innopharma Bets on Manufacturing 4.0

Pharmaceutical Technology, 2018 Solid Dosage Supplement (PDF)