Real Time Release Testing: Analytical Methods and Innovations

As part of this month's special report on real time release testing [see full story here], Pharmaceutical Technology talked to experts at analytical companies throughout the industry to gain their perspective on how RTRT can be done on a practical basis. Participating in the virtual roundtable are: Richard Godec, new product development manager, and Jonathan Yourkin, global pharmaceutical product manager, both at GE Analytical Instruments; Alon Vaisman, applications manager of pharmaceuticals for Malvern Instruments; Terry Redman, product manager for particle-system characterization at Mettler-Toledo AutoChem; and Stuart Farquharson, president & CEO of Real-Time Analyzers.

Industry implementation

PharmTech: Where does the industry now stand in terms of understanding and applying real-time release testing (RTRT) for finished drug products? Would you say that real-time release of a finished drug product is still at a nascent stage or is it further along? What are the advantages and disadvantages in real-time release of finished drug products?

Godec and Yourkin (GE Analytical Instruments): They are just beginning to study the problem. Almost no older manufacturing facilities are doing RTRT and very few new facilities are implementing RTRT for the finished products. There is more activity on in-process measurements at critical points that can help prevent final product quality failures. It is company dependent, but in general, RTRT of finished drug products is still in the nascent stage. The main advantage is the cost-effective control of the manufacturing process to meet all quality and product specifications. The main disadvantage is that there is an initial investment required to achieve RTRT of finished drug products.

Vaisman (Malvern Instruments): In my opinion, RTRT is no longer treated as an unobtainable goal. Many companies in industry and at academic research centers are making significant inroads in implementing continuous manufacturing trains and RTRT. That said, RTRT is hardly the norm today and constitutes a very small percentage of quality-control activities.

Advantages of RTRT are in streamlining production, minimizing delays and potentially eliminating out-of-spec product. The downside is that implementing RTRT can be a costly and resource intensive process that requires the thorough understanding and control of production activities.

Redman (Mettler-Toledo AutoChem): It appears that industry is making real progress in the understanding and implementation of many of the core elements of RTRT. Various PAT tools are being used earlier in development to assist the principles of quality by design (QbD), and these are facilitating the adoption of process monitoring and control concepts throughout scale-up and manufacturing. QbD enables a more thorough understanding of the relationships and interactions between critical process parameters and critical quality attributes, and this in turn allows for more effective strategies to mitigate risk and produce quality product by design. This improved process understanding, in combination with improved process analytics, is necessary for identifying the critical quality attributes and developing the methods for at-line or in-line measurement.

Although there have been successful examples of real-time monitoring and control of specific quality attributes such as granule size at the outlet of a roller compaction, or nearinfrared (NIR) spectroscopy for monitoring tablet dosage uniformity, it seems there is still a lack of holistic solutions that can reproduce or replace all of the necessary quality-control (QC) tests that would ensure a safe and effective product in its final packaged consumable form.

The main advantage of RTRT is simply business efficiency. Real-time assurance of quality, and the avoidance of storage or quarantine of intermediates and drug product while awaiting test results, will enable much greater efficiency throughout the process.

Farquharson (Real-Time Analyzers): The pharmaceutical industry clearly understands the value of RTRT. However, there is risk in applying RTRT, the greatest being incorrect information supplied by PAT tools used to implement RTRT. And consequently, I believe that RTRT is still in the nascent stage.

One of the main advantages of RTRT is being able to monitor and control the quality of a drug during manufacturing (e.g., optimizing polymorph type during synthesis, crystallization, blending). This allows the manufacturer to: improve conformity of final product to specifications; minimize the need of testing every capsule, tablet, or vial of the finished product; and identify out-of-specification batches early and eliminate further manufacturing steps (e.g., tableting of out-of-specification powder).

Technical barriers

PharmTech: What are the key technical barriers to implementing real-time release of a finished drug product overall?

Godec and Yourkin (GE Analytical Instruments): We believe the primarily barriers (not necessarily in order of priority) include the following:

• The inability of adopters to present a strong business justification to management that can justify the additional expenditures necessary to do final product RTRT. The lack of strong FDA regulatory action on non-RTRT process is probably a major cause of this. The basic business question for companies, therefore, is, Why pay the extra expense of using RTRT if a highly reliable manufacturing process can be established without it? If FDA were to require RTRT, then there would be no business doubt.

• The analytical tools for RTRT implementation are still being developed.

• Many pharmaceutical quality systems are still focused on producing pharmaceuticals using detailed standard operating procedures, such as recipes, temperatures, and timings with a three-batch process validation. It is difficult for many quality departments to allow the use of flexible process controls because they require additional subject matter knowledge that is, in many cases, unknown. Quality personnel therefore may perceive that they will have to cede control to the manufacturing department.

• There is a general lack of direct experience in RTRT projects throughout industry. Only a few pharmaceutical companies have done the required development and research to automate unit operations (processes), which is needed to implement a finished drug product RTRT system.

• Quite a few pharmaceutical companies are taking small steps in the implementation of automated controls and real-time sensors within their facilities' operations. A good example is the adoption of online total organic carbon (TOC) and online conductivity sensors for pharmaceutical waters. But even in this case, companies generally focus on the high return on investment by decreasing the number of required laboratory samples using information provided by the on-line analyzers. The additional cost of the required quality systems for real-time release of the TOC or conductivity water quality attribute is much more expensive and time consuming than can be justified with the decrease in laboratory samples. There is a new ASTM E55 standard practice, Real-time Release Testing of Pharmaceutical Water for the TOC attribute. This standard may change the number of companies that perform RTRT of their water for the TOC attribute because it addresses the technical issues involved. However, the standard does not address how to make the required quality-system changes.

Vaisman (Malvern Instruments): In-process measurements often produce results that differ slightly from QC laboratory results. Process analytical technology (PAT) equipment may be different from instrumentation used in the laboratory, and materials therefore, may often be characterized in a different state as it flows through the process. Addressing the issue of specification transfer can ease the transition of an analytical technique from development through to commercial manufacture.

In addition, RTRT often demands the successful integration of a number of analytical and manufacturing devices, a process hampered by the lack of collaboration between equipment suppliers and the existence of few applicable standards. The relatively new OPC Foundation's Analytical Device Integration (ADI) standard will help in this area but there is still work to be done.

Redman (Mettler-Toledo AutoChem): The key technical barrier I see is the integration of measurement technology with appropriate process modeling and control algorithms for holistic model-based control of the critical quality attributes (CQAS) through direct manipulation of the critical process parameters (CPPs). There is certainly room for improvement in existing measurement technologies to directly monitor CQAS and greatly enhance RTRT capabilities. However, with sufficiently reliable control of the CPPs, RTRT does not have to rely on relocating the quality assurance/quality control laboratory for in-line and at-line analysis. With true QbD, the burden of testing can theoretically be reduced to a minimal level by ensuring product quality through monitoring and control of the process itself.

Farquharson (Real-Time Analyzers): The technical barriers associated with RTRT are tied to the PAT tools being used. In addition to material property tools, spectroscopy tools, specifically Raman, NIR, and infrared analyzers, are being used to monitor the synthesis of a drug or determine the composition of a pill. Each has advantages and limitations.

For example, Raman and NIR can be interfaced into reactors via highly transmitting fiber optics, IR cannot; Raman and IR provide exceptional specificity (e.g., distinguishing polymorphs), NIR less so; NIR and IR can measure trace quantities of chemicals while Raman is typically limited to 0.1% chemical concentrations. In all cases, the variability of the measurement must be much smaller than the variability of the process. The former depends on the dynamic range and precision of the measurement (e.g., a spectral peak used to determine concentration that changes intensity significantly is better than an unresolved peak that changes very little). It also depends on the stability of the analyzer. For this reason, interferometers have dominated chemical plant installations. A shift in the x-axis, a potential problem with dispersive Raman analyzers, can result in incorrect analysis and decision making.

The key limitation for these technologies is measurements that do not correctly represent the process or product, such as the contents of a reactor or the composition of a pill.

Unit operations

PharmTech: With regard to solid-dosage manufacturing, is it now feasible to apply process analytical technology (PAT) throughout all unit operations or are certain operations not conducive to on-line or at-line testing?

Vaisman (Malvern Instruments): Not only is it possible, it is being done. A wide variety of PAT equipment is already commercially available for monitoring, for example: homogeneity, composition, particle-size distribution, density, moisture content, and other parameters. These technologies are being applied across the spectrum of unit operations used by the pharmaceutical industry. While certain applications are particularly exacting in terms of the demands they place on an analytical technique, I believe progress is being made in almost all relevant areas.

Redman (Mettler-Toledo AutoChem): PAT has applications in every unit operation, and is achievable with existing measurement technologies. It is not always feasible or cost-effective to implement direct measurement of all CPPs and CQAs. In many cases, gaps in measurement technology must be filled with inferred measurements that must be proven statistically reliable.

We expect to see improvement in measurement technologies across the board though, which will continue to improve the ability to monitor for process control purposes and provide analytical measurements for quality control.

PAT applications in solid dosage

PharmTech: Can you offer some specific examples of PAT being applied in solid-dosage manufacturing? How would the data used at this particular step be used for the real-time release of a finished solid-dosage product?

Vaisman (Malvern Instruments): A common unit operation in the pharmaceutical sector is milling, to produce actives and excipients of defined particle-size distribution for subsequent processing, and/or formulation. On-line particle size analysis enables the automatic control of a mill and provides continuous monitoring of the exiting material.

In "A PAT Solution for Automated Mill Control," published in Pharmaceutical Technology's January 2010 issue, the authors describe the use of on-line laser diffraction particle-size measurement to automate mill control at a commercial site. The generic solution developed has widespread application. With automated control in place, the operator selects a particle size set point for the product. The speed of the mill rotor then varies automatically in response to real-time particle-size data, so that the specification of the exiting material is consistently maintained, even during fluctuations in the feed. The result is better product quality, higher throughput, and less waste.

This solution could be directly applied for RTRT provided that correlations between laboratory QC and the process instrument are in place. This is a readily achievable goal with laser diffraction particle-size analysis.

Redman (Mettler-Toledo AutoChem): In solid-dosage manufacturing, the size distributions of particles and granules are known to be critical parameters affecting final product quality (i.e., influencing critical powder properties such as compressibility and flowability, which in turn impact the tableting step that determines final product quality attributes such as dissolution rates and bioavailability.) Recently, in-line techniques, such as focused beam reflectance measurement (FBRM) and at-line techniques such as laser diffraction, have been shown to provide the capability to measure the progress of size enlargement processes by roller compaction, high-shear wet granulation, and fluid-bed granulation. The ability to monitor particles in real time permits the control of continuous roller-compaction processes through the manipulation of roller speed, gap size, and roller pressure. Likewise, batch processes, such as high-shear or fluid-bed granulation can be monitored until the granule distribution hits a predetermined endpoint. In either case, measuring the particles in-line or at-line provides the ability to control the granule-size distribution.

Farquharson (Real-Time Analyzers): We have used Raman spectroscopy to monitor drug synthesis and crystallization (kinetics and yield) in-line, and polymorphism at-line, steps that precede compaction, tableting, and so forth. We have also used chemometrics, full spectral analysis, to relate process parameters to product quality (e.g., product to by-product ratio). The greatest advantage of RTRT is the ability to identify when a process/manufacturing step is heading out-of-control so that corrective action can be taken. Even if an out-of-specification product is made, savings can be realized by not continuing with the remaining steps. For example, blending an out-of-specification active (e.g., too much byproduct) with the incipient, followed by compaction, coating, and so forth, can be a substantial financial waste.

Changing analytical methods

PharmTech: How does analytical-methods development change when working in a PAT environment? What are some of the challenges that might arise in scaling up a process from a laboratory scale to the manufacturing scale?

Godec and Yourkin (GE Analytical Instruments): Analytical methods, if properly validated to the requirements of the International Conference on Harmonization Q2(R1) guideline on analytical validation, and for their intended use, will not generally change. Analytical methods can be subject to scaling problems, however. For example, scaling problems can occur when the automatically collected or measured sample is not representative of the bulk-material analyte concentration.

Vaisman (Malvern Instruments): Method development for a PAT system must take into account a wider scope of parameters than just the measurement system and material. In particular, the process interface must be carefully considered to ensure representative analysis. Techniques where results can be affected by multiple variables tend to cause more difficulties during scale up, and in many cases, a new method must be developed for larger scale.

For example, contrast the example of particle-size analysis by laser diffraction and by ultrasonic extinction. The former technique is based on first principles and is relatively insensitive to variations in environmental factors. Laser-diffraction results will change only if the particle size of the product changes. On the other hand, the results produced by ultrasonic extinction are dependent on multiple factors, including temperature, moisture, bulk density, and others. Should any of these factors change during scale up, the result will be affected even if the monitored value (particle size) has stayed the same.

Farquharson (Real-Time Analyzers): The greatest challenges in scaling up a process is maintaining yield and minimizing byproducts. This difficulty is largely due to the fact that mixing and heat are not uniform at larger scales. Consequently, a laboratory-based analytical method may not be easily transferred to the process. For this reason, Raman and IR may be preferred over NIR because of their greater information content, and because the analysis may need to be changed. If complex models are used, such as chemometrics, model transfer from analyzer-to-analyzer may be an issue. X-axis stability and resolution must be maintained. The latter is problematic if a dispersive analyzer is planned for the process, but an interferometer-based Raman analyzer was used in the laboratory. It is also worth stating that full-scale reactors are not always as pristine as laboratory-scale reactors, and fluorescence interference can occur when using 785 nm lasers to generate Raman.

Sampling plans

PharmTech: How should sampling plans be approached when using PAT and RTRT compared with traditional sampling methods?

Vaisman (Malvern Instruments): Usually, PAT instruments access larger amounts of sample than would be used for laboratory analysis. On-line instrumentation would typically use automated sampling, in contrast to traditional off-line techniques that tend to rely on grab-sampling at the end of the process. Sample preparation is usually relatively limited for PAT systems to achieve the measurement rates required for continuous monitoring.

Farquharson (Real-Time Analyzers): The best place to monitor reactors is in-situ (i.e., inside the reactor). This can be accomplished using a long-rod fiber-optic probe. However, efforts must be made to keep the probe head clean. In-situ measurements allow for true real-time monitoring. This is significantly better than traditional grab-sampling, which is primarily used to determine whether the end-point had been reached. This, of course, can lead to incorrect analysis and action, specifically if the reaction is running hot or cold (i.e., high yield or low yield).

Equipment failures

PharmTech: What is the best practice to date for handling an equipment or analytical instrumentation failure during RTRT?

Godec and Yourkin (GE Analytical Instruments): The implementation of analytical backup procedures is a requirement, and can be achieved by using redundant equipment or backup laboratory procedures.

Farquharson (Real-Time Analyzers): The best approach is to have internal analyzer diagnostics. For example, we use a process Raman analyzer that constantly monitors a number of parameters, such as the laser power and the unit temperature. It is also possible to implement diagnostics to avoid other failures, such as issuing a warning if the analyzer indicates a constant value due to a coating forming on the optical interface.

Advances in instrumentation

PharmTech: What have been some recent advances in analytical instrumentation to facilitate PAT implementation in solid-dosage manufacturing? Also, what would be helpful for further PAT implementation?

Godec and Yourkin (GE Analytical Instruments): In solid-dosage manufacturing, NIR has benefitted customers trying to understand variation in the process and to improve process understanding. The holy grail with any PAT implementation is the ability for pharmaceutical manufacturers to now have the ability to monitor and control critical processes to achieve product and process robustness. The tool for this achievement would be accurate and reliable on-line analyzers, regardless of the method.

Vaisman (Malvern Instruments): One relatively recent advance is the development of spatial filter velocimetry for high-shear granulation.

Probes that use spatial filter velocimetry can be installed directly in the granulator to provide real-time measurement of the growing granulate. Working across the size range 50 to 6000 microns, such probes are sufficiently robust to measure accurately in the damp, sticky conditions that exist within the vessel. The data provided is extremely useful for endpoint detection, a key processing requirement.

Looking ahead, accurate and reliable in-line measurement of blend homogeneity and the continuous monitoring of segregation, during roll compaction, and in the tablet press, remain important long-term goals.

Redman (Mettler-Toledo AutoChem): Process analytical companies continue to innovate for specific unit operations as challenges arise. Real-time monitoring of granule size in a high-shear wet granulation step is especially challenging, as the sticky mass is difficult to sample and adheres to any instrumentation within the process equipment. The process also occurs so quickly that off-line sampling and analysis is impractical. Mettler Toledo has developed hardware and software solutions to enable probe-based FBRM technology to measure the granulation process in real-time. A hardware-based mechanical scraper uses an intermittent wiping mechanism to remove granulated product that sticks to the probe window. On the software side, advancements in digital signal processing allow the FBRM system to detect particles and granules that are physically stuck to the window. These stuck particles are then eliminated from the measurement to ensure the integrity of the real-time process data.

For the future, it is important that industry continue to develop best practices by collaborating with regulatory agencies, consultants, and process analytical technology suppliers. The wide range of regulatory roadblocks and business incentives continue to make RTRT and PAT overall a confusing route. Greater clarity of the current situation and the path forward will greatly accelerate development and implementation of solutions.

Farquharson (Real-Time Analyzers): Raman, FT-IR, and NIR have great present and future value to PAT. The development of process-worthy FT-Raman spectrometers can overcome the limitation of dispersive systems currently in use (e.g., X-axis instability and fluorescence interference).

Although, large spot and transmission Raman have improved representative analysis of capsules, pills, and tablets, analysis is still inaccurate and/or slow (i.e., both modifications reduce sensitivity). Future designs may alleviate this problem.

Be sure to check out the additional features related to this article on powder processing and on industry and regulatory expectations