The ability of continuous manufacturing to eliminate the need for process scale up during drug development may be the greatest prize for makers of pharmaceutical products.
Many industries, most notably the bulk chemicals sector, have long since switched the majority of their production away from batch manufacture to continuous processing in order to capitalize on well-documented gains in manufacturing efficiency, capex and labor utilization, as well as product consistency. This transition has been pivotal in streamlining the supply chain for many core commodities, while simultaneously reducing costs—an important societal gain. In contrast, and as noted by the FDA1, “manufacturing experts from the 1950s would easily recognize the pharmaceutical manufacturing processes of today.” However, that situation is now changing fast, with intensive effort going into the development of continuous manufacturing (CM) processes within the pharmaceutical industry.
Over the next decade or so, these processes are expected to transform the long-term manufacturing efficiency of the industry, but there is arguably an even greater driver for change. The unique regulatory environment associated with drug development makes process scale-up complex, technically challenging, and labor and material intensive. One of the most compelling benefits of CM for the pharmaceutical industry is, in fact, proving to be its ability to eliminate the scale-up step and permit the robust scoping of a commercial design space early in the development process.
In this article, we review the potential benefits of CM for this sector, focusing on tablet production, assess progress to date, and consider the technology required for automation and real-time release (RTR). Particle sizing technology is used to exemplify the integration of real-time monitors within a continuous process to exert the necessary, automated control and ensure consistent product quality.
Batch vs. continuous manufacture
In a true batch process, raw materials are added at the start of the processing period and the product is discharged at the end. No material is added or removed at any other time. The state of the in-process material therefore changes with time. In a continuous process, in contrast, raw materials are continually added, product is continually discharged and the process operates at a steady state. This stable operational state makes it far easier to achieve high levels of product consistency2.
Continuous Manufacturing: Uniquely Valuable for Pharma?
Though CM has well-documented generic benefits, it is becoming clear that it also has the potential to address some very specific issues for the pharmaceutical industry.
Traditionally, tablet production involves a series of individual batch unit operations, operated over a period of weeks or even months, often on multiple sites. Synthesis of the Active Pharmaceutical Ingredient (API) is typically a completely separate process from tablet manufacture. Figure 1 contrasts integrated continuous tablet production processes, running from synthesis of the API through to tablet production, though blending of the API and excipients as the starting point is an alternative, potentially more manageable, option.
Assuming both processes are effectively controlled, the continuous route directly tackles a number of healthcare and pharmaceutical manufacturing concerns by:
More reliably delivering consistent product
Drug shortages are a well-documented issue in the USA, with sterile injectables such as chemotherapy, anesthesia, and other acute drugs particularly susceptible to disrupted supply. The FDA has invested considerable resource in addressing such shortages, and has identified poor product quality arising from manufacturing issues as a key source of supply disruption3,4; the two other major causes are quality issues arising from delays/capacity and raw material supply limitations. Against this backdrop, more reliable and consistent manufacture is an important goal.
Streamlining the supply chain
Eliminating discrete steps in the supply chain, and thus also reducing the associated potential for hold-ups, is particularly advantageous when dealing with APIs and/or excipients that are prone to degradation or sensitive to environmental conditions. For these products, streamlined production has the potential to improve overall product quality. More generally, a shorter supply chain is less prone to disruption, easing the smooth flow of drug supplies.
For an equivalent production rate, a continuous process utilizes a smaller equipment footprint by delivering a higher throughput per unit volume, per unit time. CM therefore offers opportunities for reducing both capital and operating expenses and ultimately, the cost of drug manufacture.
Reducing material inventories
Smaller inventories are associated with enhanced manufacturing safety, especially when dealing with highly energetic or otherwise hazardous materials.
Scale-up in its traditional form—the progressive increase of the volume of processing equipment—is not relevant in a CM environment, since production volumes can be increased simply by operating the unit for longer campaigns, increasing throughput and/or installing a parallel line. Typically, production rates become a function of operating time rather than operating volume. This means that a continuous unit offers far greater flexibility in terms of scale of operation than a batch equivalent. This flexibility makes it possible to:
- Ramp up production to meet healthcare demands, for example, in the event of a pandemic
- Tailor production rates to meet the needs of specific geographies, allowing the needs of individual countries to be met locally, with the same processing equipment
- Accelerate and enhance process development.
With time to market a crucial determinant of profitability, it is actually this very last gain that is arguably the greatest prize.
Cutting the Time and Cost of Process Development
A QbD approach to process development demands a rigorous understanding of the links between the critical quality attributes (CQAs) of the pharmaceutical product—those that influence its clinical efficacy—the critical material attributes (CMAs) of the raw materials, and the critical processing parameters (CPPs) applied during manufacture. Unfortunately, for many processes, these relationships can be scale-dependent. Consequently, with a traditional batch approach, as the scale of production is increased towards commercialization, the design space for the process shifts. This effect complicates both the submission process and the provision of material for clinical trials, which must be rigorously representative of ‘as manufactured’ material.
In contrast, if a new drug submission is prepared using a CM process, then it is possible to carry out experiments on a commercial scale unit, simply by operating the process for a much shorter period than will be required for commercial production. This is a major gain. It means that the formulation can be finalized early in the development process to produce representative material for clinical trials, and that the design space is specified using commercial scale equipment.
Furthermore, experimentation with a continuous, closely-monitored process is very efficient. Such units operate in a steady state, making them more amenable to process control and more responsive to change than batch units, which in contrast operate in a transient state. With a continuous process, it is far easier to identify and understand correlations between process and raw material changes and CQAs. CM therefore allows representative and robust scoping of the design space, and the associated development of predictive models, using less API than would be required with a batch train. Indeed, published estimates suggest that savings in API associated with the development of a wet granulation process could be in the region of 75 percent4.
Tackling the Challenges of CM
These drivers provide an impetus to address key challenges associated with the implementation of CM. One such challenge is the definition of a batch, but the greater, not unconnected demand, is to achieve adequate process monitoring and automated control.
The issue of batch definition is relatively straightforward if a CM process is operating at a well-controlled steady state, with material progressing from one unit operation to the next sequentially, in ‘plug’ or mass flow; i.e., material is not back-mixing from one unit operation to another. Provided these conditions are met, a batch of product of consistent quality can be defined based on time, even if the inputs to the process have varied during that period. However, if the process is not well-controlled, the concept of defining batch based on time begins to break down. Achieving effective process control is therefore crucial for the implementation of CM.
The FDA references three different levels of process control—Levels 1 to 3—which are associated with increasing complexity and responsiveness6 (see Figure 2). Many batch pharmaceutical processes operate under Level 3 control, relying on end point testing after each unit operation to verify that the product is suitable for subsequent processing. This level of control requires only limited understanding of the links between material inputs and processing parameters and the product quality, relying instead on the application of very tightly constrained operating conditions. Level 2 control offers greater operational flexibility within a well-defined design space and is more closely aligned with the application of QbD, which defines suitable ranges for identified CMAs and CPPs. Level 1 similarly demands extensive understanding of the product and process and implies real-time automatic process control based on the implementation of appropriate Process Analytical Technology (PAT) for process monitoring.
Implementing CM with Level 3 control is almost always impractical, since end product testing after each unit operation is inconsistent with an integrated, continuously operating process. Level 2, or preferably Level 1, control is therefore typically the goal, making the identification of appropriate PAT inextricably linked with the successful implementation of CM.
A Commercial CM for Tablets
Figure 3 shows a schematic of a commercialized CM unit for tablet production6 and identifies a number of PATs with proven application for this process. These include:
- In-line Near Infrared Spectroscopy (NIR)—for compositional analysis, including API (potency) confirmation, and water content measurement
- Loss-in-weight (LIW) feeders—for real-time mass flow measurement to monitor and control the dosing of individual tablet components
- In-line laser diffraction particle size analysis—for granule size measurement, ahead of tableting
- Tablet tester with integrated Raman probe—for verifying the CQAs of the finished tablets including: weight, thickness, hardness and the physical form of the API
The raw feeds to the tableting process are under mass flow control (LIW feeders) (Step 1). The resulting blend is granulated (Step 2) and then dried, meeting an exit temperature set point (Step 3). The dried granules are milled (Step 4) and water content is subsequently verified by NIR, while particle size is controlled based on laser diffraction particle size measurements. Further ingredients are added to the granules under mass flow control (Step 5), ahead of compression (Step 6), and the properties of the finished core tablets are then analyzed using a tablet tester incorporating a Raman spectroscopic probe, which verifies the physical form of the API in the finished dose. Coating is the final step (Step 7).
PAT is an integral part of this process, and the data it generates is used for In-Process Control (IPC) and/or for RTR Testing (RTRT). Measurements associated with IPC are used to assess material against operational set points and provide a warning of non-conformance. The outcome of an IPC measurement may result in an automated adjustment to the process, or initiate segregation of the in-process material from the process. A prime example of an IPC measurement is LIW feeder flow, which is used to control the feed of individual components to the blenders.
When in-process material drifts out of specification, it must be removed from the process so that the integrity of the product batch is assured. In the process above, segregation is possible after the drying and milling step and/or after compression, and is instigated as early in the process as possible once results indicate it is required. In summary, PAT keeps the in-process material within the required specification at all times, or, if an operational issue does develop, detects any deviation in a timely way to ensure the removal of out-of-specification material from the process.
PAT analysis of in-process material can also be leveraged for RTRT if it appropriately characterizes the CQAs of the product. Table 1 lists typical CQAs for a tablet and shows how they can be assessed at various points in the process. For example, dissolution, a CQA because of its impact on bioavailability, can be assessed by measuring the particle size of milled granules (Step 4 in the process above), API and water content in the final blend (Step 5) and the weight, thickness and hardness of the finished tablet (Step 6). Generally speaking, PAT measures much larger sample sizes than are routinely sampled in endpoint testing and, when appropriately implemented, can provide a robust assessment of finished product quality. Tablets may therefore be released in accordance with the test schedule shown in Table 1, but a final composite sample is usually taken for retains, stability testing and any release testing not amenable to in-process implementation, such as appearance.
Implementing PAT: Particle Sizing
The implementation of each PAT as part of the overall CM process is a significant exercise, with method development and validation a key focus. Looking at particle sizing as an example helps to elucidate the features of a technique that make it particularly amenable to in-process implementation and the steps involved in integrating a PAT within the process.
Laser diffraction systems exploit fundamental aspects of the behavior of light, as mathematically described by Mie theory. Particles illuminated by a collimated laser beam scatter light over a range of angles, with larger particles scattering at relatively high intensity at narrow angles to the incident beam, and smaller particles producing a lower intensity signal at much wider angles. Laser diffraction analyzers detect the scattered light produced by a given sample and apply Mie theory to calculate the particle size distribution responsible for it.
Laser diffraction is a fast, non-destructive technique that is highly amenable to automation. Modern systems are readily installed in a wide variety of process equipment—inline or online, on pipework or sampling from a vessel—and have been refined for easy integration within multivariate process control systems. However, ensuring a robust interface between the process and the analyzer is critical to the success of a continuous particle sizing installation.
Figure 4 shows an in-line sampling system used to measure the size of particles exiting the cyclone separation step of a pharmaceutical process. The vortex breaker induces linear flow in its shadow and the sampling system design has been optimized based on empirical testing. It covers 25 percent of the cross-sectional area of the pipe, meaning that 25 percent of the production run is sampled. In comparative tests, this set-up gave the best correlation with offline, reference particle size measurements, proving it superior to alternative commercially available sampling systems. Analogous set-ups can be used to sample CM processes as and where required, such as post-milling—as indicated in the process shown in Figure 3.
Figure 5 shows the results of a test of the inline analyzer. Based on a thorough understanding of the process, a change is made to induce a three-micron increase in the particle size, Dv50 (the particle size below which 50 percent of the particle population lies based on volume) of the measured granules. The particle size analyzer records a pre-change Dv50 of 74.2 µm and a post-change value of 77.1 µm, responding immediately to the change and stabilizing after a period of around eight minutes, because of the averaging algorithm applied. Tests such as these give a high degree of confidence in the inline data, which is further boosted by ongoing comparisons with offline measurements (see Figure 5b).
Generating representative data is crucial but, when it comes to automated control and CM, the ease with which the data can be used as part of an automated process control system is equally critical. Figure 6 shows the full software and hardware architecture associated with the particle size analyzer. The raw scatter pattern measured by the analyzer is converted to particle size distribution data by a dedicated PC (Malvern PC). Process control is based on an 8-minute block average Dv50; averaging being a useful strategy for damping the fluctuations that can be observed with individual measurements. Both raw data and process data are archived and the entire system is subjected to continued method verification to confirm the validity of the data being produced.
Exploring the Benefits of CM in Process Development
Using a CM unit for process development allows a three-stage approach based on Design of Experiment (DOE) studies of individual unit operations, of relevant multiple unit operations grouped together, and of the complete process. This makes it possible to define operating ranges for specific units, but then refine these ranges based on an understanding of interactions between the units and the more effective linking of CMAs and CPPs with CQAs.
Figure 7 shows data from a trial of a single unit, a ConsiGma-1 Twin Screw Granulator that illustrates the speed with which DOE can be carried out and the minimal API required when appropriate PAT is in place. A total of 27 experiments were performed to assess the effect of screw speed, water content and percent (%) binder on the Dv50 of the granules produced. The unit was run for one to four minutes to capture data for each operational point, with one minute gaps between experiments to allow the unit to reach steady state. A total of 16 kg of API was used for the complete trial.
Figure 8 shows the results from the extension of this trial to a more advanced study designed to assess the impact of changes in granule properties on the performance of the fluidized bed dryer. Such studies exemplify the relative ease of design space scoping with a CM, and illustrate how the data gathered can be used to provide a foundation for process model development. The final step in such studies is to extend them to the entire line to build a robust design space for the process, scoped on commercial scale equipment.
The benefits of CM are well-proven by, for example, the bulk chemicals sector, but for the pharmaceutical industry, a switch from batch to continuous processing has the potential to address certain specific issues. More consistent product quality, a streamlined supply chain and easier scaling to meet local needs, or in the event of a pandemic, are major potential gains. However, it is the ability of CM to eliminate the need for process scale up during drug development that may be the greatest prize.
Developing new drugs using CM processes eases access to representative product for clinical trials and, at the same time, enables submission based on design space specified at a commercial scale. Better, faster process development, with less API, is therefore part of the potent case that can be made for CM. The commercialization of rigs for tablet production demonstrates what can be achieved and at the same time highlights the pivotal role of PAT in enabling the monitoring and control required for effective continuous processing. Over the near-term, such units have an important role to play in transforming not just the efficiency of pharmaceutical manufacture, but of drug development, too.
1 Chatterjee, S. FDA Perspective on Continuous Manufacturing. Presented at IFPAC annual meeting 2012. Available at: https://www.fda.gov/downloads/AboutFDA/CentersOffices/OfficeofMedicalProductsandTobacco/CDER/UCM341197.pdf
2 Lee, S et al. ‘Modernizing Pharmaceutical Manufacturing: From Batch to Continuous Production.’
3 ‘Frequently asked questions about drug shortages’ – available to view at: https://www.fda.gov/Drugs/DrugSafety/DrugShortages/ucm050796.htm#q3
4 Drug shortages. FDA. 2016. Available at: https://www.fda.gov/downloads/Drugs/DrugSafety/DrugShortages/UCM441583.pdf
5Thomas, H ‘Utilizing continuous processing for streamlines and efficient development of QbD products’ Presentation delivered at IFPAC 2012.
6 Warman, M et al. ‘PAT for In Process Control and Real Time Release Testing in Continuous Manufacture’ Presented at APACT 2016.
7Warman, M. Expected vs Predicted, parametric Release v RTR. Presented at IFPAC 2011.
About the Authors:
Martin Warman is Company Director at Martin Warman Consultancy and Alon Vaisman is Head of Sensors and Automation Solutions at Malvern PANalytical.