`
`research focus
`
`reviews
`
`Advances in pharmaceutical materials
`and processing
`Kenneth R. Morris, Steven L. Nail, Garnet E. Peck, Stephen R. Byrn,
`Ulrich J. Griesser, Joseph G. Stowell, Sung-Joo Hwang and Kinam Park
`
`Advances in pharmaceutical materials and processing require new
`
`generations of pharmaceutical technologies, which in turn require an
`
`improved understanding of each step in the unit processes of dosage
`
`form development. The unit processes range from raw material
`
`qualification to final product release using process monitoring of
`
`critical steps. The authors illustrate some recent research trends in
`
`understanding and improving pharmaceutical materials and process-
`
`ing through the use of experience obtained within several research
`
`programs at Purdue University (West Lafayette, IN, USA).
`
`Kenneth R. Morris
`Steven L. Nail
`Garnet E. Peck
`Stephen R. Byrn
`Ulrich J. Griesser
`Joseph G. Stowell
`Sung-Joo Hwang
`and Kinam Park*
`Purdue University
`School of Pharmacy
`West Lafayette
`IN 47907, USA
`*tel: 11 765 494 7759
`fax: 11 765 496 1903
`e-mail:
`esp@omni.cc.purdue.edu
`
`t In common with other industries, the phar-
`maceutical community is preparing itself for ad-
`vancing into the year 2000 and beyond. Many
`innovations are taking place, both in the devel-
`opment of new drug delivery systems and the
`project plans for new drug development activ-
`ities. All of these areas are important factors in
`the future of new drug products in any attempt
`to relieve current or future ailments.
`
`Next generation of pharmaceutical
`technologies
`As technology progresses, pharmaceutical scien-
`tists must also move forward to develop improved
`treatments for both old and new afflictions. An
`area of particular importance is drug delivery.
`One of the current activities taking place in this
`area is site selection for drug absorption in the
`gastrointestinal tract; possible sites may include
`the buccal cavity, stomach, small and large intes-
`tine and the rectum. Traditionally, drug delivery
`has been achieved through the small intestinal
`route, but now other areas of the body are being
`explored. This movement into advancing drug
`delivery will impact upon methods used in the
`
`preparation and control of dosage forms, and thus
`new methodologies may be required for the
`development of the dosage forms of the future.
`
`Compliance issues
`In addition to these considerations are those that
`involve the US Food and Drug Administration
`(FDA) and its expectations regarding the
`documentation of new drug development activi-
`ties. The agency’s interest in the basic formu-
`lation development and technology transfer is-
`sues has increased during its pre-approval and
`post-approval inspections. These considerations
`are important in the design of both the processes
`and equipment of the future. Pharmaceutical sci-
`entists must consider what the agency may re-
`quire in these areas. It is important to be aware
`of the fact that FDA pays a great deal of attention
`to vendor sites and laboratories that may be listed
`in a new product approval document. At present,
`a great deal of emphasis is placed on current good
`manufacturing practices (cGMPs) and how they
`are committed to new drug applications (NDAs)
`and abbreviated new drug applications (ANDAs).
`In terms of clinical trials, documentation and
`records are important factors in the support of a
`new product’s research and development (R&D),
`scale-up and commercial production.
`
`Analysis of unit operations
`Because unit operations may have to be modified
`to suit the requirements of a new dosage form,
`developers’ activites in scale-up and production
`will be carefully scrutinized and thus they must
`be aware of their processing and equipment
`needs. As demonstrated in Table 1, the origins of
`the activity centers around raw material qualifi-
`cation and continuous efforts to improve specifi-
`cations, which are often beyond those used in
`
`Copyright ©1998 Elsevier Science Ltd. All rights reserved. 1461-5347/98/$19.00. PII: S1461-5347(98)00062-5
`
`235
`
`Merck Exhibit 2163, Page 1
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`
`
`reviews
`
`research focus
`
`PSTT Vol. 1, No. 6 September 1998
`
`Table 1. Current considerations important to future development of the unit operations for pharmaceuticals
`
`Unit operation/activity
`
`Current status
`
`Next level
`
`Cutting edge
`
`Raw material qualification
`
`Drying of hydrates/solvates
`
`Mixing/blending
`
`US Pharmaceutical specification
`and methods
`Internal process specific
`specification
`
`Trial and error determination of
`drying cycle
`Variety of blender types
`monitoring by thief sampling
`
`Dry granulation
`
`High shear blending/slugging
`
`Wet granulation
`
`Drying
`
`Compression
`
`Lyophylization
`
`End-product testing
`Packaging
`
`Process control
`
`High shear and fluid bed
`granulators (endpoint
`determination by predetermined
`time and/or periodic sampling)
`
`Tray, fluid bed drying
`
`Rotary tablet presses
`Compression force or thickness
`priority
`Empirically determined cycles
`
`Dissolution, hardness
`Semi-automated lines
`Spot inspection
`Minimum sampling
`
`Product containment
`
`Limited equipment
`
`Improved internal specification
`including more physical parameters
`(e.g. shape, surface roughness)
`
`Full solid-state characterization of
`components during cycle
`Bin or high shear blenders
`Remote monitoring for endpoint
`determination
`Roller compaction
`Uniformity monitoring
`
`Vacuum granulator/dryer
`combinations (e.g. Zanchetta)
`Extrusion/sphearanization
`Remote monitoring of granulation
`endpoint
`Standard sampling and testing for
`water
`Fully instrumented presses
`
`Predictive methods to correlate
`physical properties to observed
`behavior (e.g. crystal morphology,
`fractal analysis of surface
`roughness)
`Control of forms during drying
`
`Feedback control of process by
`monitoring device
`
`Full process control through
`optimization
`
`Control of granule charactistics by
`real time modification of process
`
`Near infrared methods
`
`Feedback-controlled presses
`
`Cycle optimization using advanced
`analytical techniques
`Sampling
`100% inspection
`
`‘Smart’ freeze dryer
`
`Final release by near infrared
`
`Computer data processing
`improvement
`Improvement in container design
`
`Parametric release
`
`Islands of operation
`
`the United States Pharmacopeia (USP). Specifications should
`become dynamic, and factors such as particle shape or surface
`roughness have to be considered in attempts to improve the
`processing of a particular material. It is important to note that
`product formulation involves the bringing together of particles
`of different surface morphology and this may impact upon any
`further processing. In the preparation of various dosage forms,
`and solids in particular, there are always concerns with the
`blending of dissimilar materials. These concerns involve not
`only the blender type, but also the fundamental operational
`characteristics of the blending device. It is therefore necessary
`to look to the future for systems that may be able to offer a
`form of feedback control of the blending process and this will
`require appropriate monitoring of the system.
`One unit operation that is considered more frequently is the
`preparation of granules for tableting. These are prepared by
`
`either dry granulation techniques (using roller compaction) or
`wet granulation. Problems associated with wet granulation may
`make it necessary to adopt the dry granulation technique in
`order to achieve adequate control of the distribution of the ac-
`tive ingredients. Advances have been made over the last ten
`years, although there is often inadequate control over the gran-
`ule characteristics, and there is a need to optimize the process
`of wet granule formation. This will require certain modifi-
`cations of the process, or a second look at methodologies that
`have been discarded. The monitoring of the endpoint of wet
`granulation is still a technique of concern and various methods
`of monitoring the endpoint of wetting require examination.
`There is interest in the general monitoring of drying, opti-
`mization of tablet machines in current use, and investigations
`into freeze-drying techniques for improved or optimal product
`preparation methods.
`
`236
`
`Merck Exhibit 2163, Page 2
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`
`
`PSTT Vol. 1, No. 6 September 1998
`
`research focus
`
`reviews
`
`In-line and on-line control
`In an analysis of the various pharmaceutical unit operations, it is
`possible to identify some in-process control parameters that can
`be monitored so that the collected data can be fed into a com-
`puter.This will allow release of the product by so-called ‘paramet-
`ric’ procedures. Parametric release presents a challenge to those
`involved in solid dosage forms because of the limited amount of
`techniques currently available for in-process monitoring.
`Of current concern is the need for containment, which is
`necessary for worker protection when the active drug is a
`highly potent material. Containment allows the manufacturer
`to transfer materials during various manufacturing steps, in-
`cluding in-process manipulation of materials and then final
`transport to a tablet machine area or a capsule filling area with-
`out production personnel contact. The manufacturing plant of
`the future will contain isolated areas, which are sometimes re-
`ferred to as ‘islands of automation’. These are sub-units within
`a process design for the purposes of containment. Certain types
`of drug substance may frequently require manufacturers to be
`completely enclosed in appropriate suits with connecting air
`supplies, and thus it may be possible to work within these is-
`lands of automation and eliminate risk of worker exposure to
`hazardous substances. The authors have experience with ‘lights
`out’ operation, and within tablet manufacture in particular.
`This method is a further design element that would require
`decreased levels of personnel contact with the material. The
`pharmaceutical processing plants of the future will need to
`have rigorous containment capabilities.
`Purdue University has been recognizing the need to improve
`both pharmaceutical processing and the characterization of
`pharmaceutical materials. As part of these efforts, Purdue
`University has established the National Science Foundation
`(NSF) Industry/University Cooperative Research Center in
`Pharmaceutical Processing, and, with the Massachusetts Institute
`of Technology (MIT), the Consortium for Advanced Manufac-
`turing of Pharmaceuticals (CAMP).Through the NSF Center and
`CAMP, faculty members associated with the programs are in-
`volved in cutting edge research in pharmaceutical materials and
`processing.This article describes several projects at Purdue Uni-
`versity that are supported by the two programs. In specific
`terms, strategies for the identification and control of processing
`variables in each unit operation of the pharmaceutical manufac-
`turing process are described. There are other promising ap-
`proaches under development which are not addressed here.
`
`Physical properties in raw material qualification
`Raw material qualification is an important part of process vali-
`dation and it is becoming increasingly important as economic
`reasons lead to an increase in levels of outsourcing of the
`manufacture of active compounds. Variation in raw materials
`
`Box 1. Issues in qualification of physical
`properties
`• Lot-to-lot and batch-to-batch variation
`• Variation in material from alternate supplier
`• Representative sampling establishing meaningful specification
`• Scaling results
`• Appropriate characterization methods
`
`can lead to failures in production and/or dosage form perfor-
`mance, which are often attributed to an uncontrolled process.
`Conversely, a process is not in control if it is not ‘rugged’
`enough to be able to accommodate the normal range in vari-
`ation of the physico-chemical properties of its components.
`Some of the major issues surrounding physical property quali-
`fication are listed in Box 1.
`Lot-to-lot and batch-to-batch variation occurs because the
`raw material supplied by either an internal or external source is
`also the result of a process with its own intrinsic variations.The
`limits of variation should be defined for a controlled process;
`however, suppliers cannot be expected to control those proper-
`ties that are of differing levels of importance to each manu-
`facturer. Since multiple suppliers of a component are typically
`identified in the NDA, ANDA or supplementary new drug
`application (SNDA) stage, the variation in materials received
`from different suppliers must be examined carefully. A major
`problem is representative sampling; that is, in attempting to ob-
`tain analytical results on a small sample that represents the true
`characteristics of the bulk product. In terms of difficulty, this is
`second only to trying to obtain a representative sub-sample.
`The results of lab- and/or pilot-scale testing can have a variable
`relationship to behavior at levels at the production scale.There-
`fore, the range of acceptable variation must be carefully estab-
`lished prior to technology transfer. It may be necessary to stress
`the limits by producing batches of product with the selected
`raw material that are at the extremes of the property limits.
`However, attempts to fail batches intentionally is not easily
`done or indeed justifiable above the pilot scale. All of these is-
`sues assume that there is a method to measure key physical
`properties and that they may be correlated to the behavior of
`the material in the unit operation of impact. Many methods
`have, and are, being developed1, although the traditional ana-
`lytical methods are often insufficient for the development of a
`correlation with performance. A number of methods are cur-
`rently under development at Purdue University, and two of
`these methods are discussed here.
`
`Method for indexing crystals in morphology determination
`Determination of the effect of crystal morphology on how the
`material will behave during handling is difficult to achieve.
`
`237
`
`Merck Exhibit 2163, Page 3
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`
`
`reviews
`
`research focus
`
`PSTT Vol. 1, No. 6 September 1998
`
`Particle size, charging and morphology may interact, making it
`difficult to factor the contribution of each property to the ob-
`served behaviors.While there are reproducible methods for the
`measurement of size, for several reasons morphology is more
`difficult to measure. None of the current methods address the
`issue of determination of the crystal habit; that is, the crystal-
`lographic faces that are exhibited by the crystal. Morphology
`measurements made without this information ignore the
`possible differential characteristics of distinct faces.
`Crystal faces have traditionally been indexed on one crystal
`that must be of relatively high quality.The methods of choice are
`optical goniometry, which determines interfacial angles, or ob-
`taining the reflection on a single crystal unit with subsequent
`comparison to the known single crystal structure2. The method
`used by the authors also requires knowledge of the single crys-
`tal structure; however, it is performed on a number of crystal-
`lites of varying quality. In addition, the method allows for the
`indexing of fragments of crystals, which may then be used to re-
`construct the original morphology. This ‘reverse’ crystal engi-
`neering concept is in recognition of the fact that no matter how
`carefully one controls the morphology of the bulk drug, subse-
`quent unit processes have the last word. An example that em-
`ploys acetaminophen is described here, and it demonstrates
`both the utility and ease of the method. It is a simple way of de-
`termining the indices of crystal faces, and, because it is used on
`multiple crystals, the level of certainty of the index is high.This
`method is equally applicable to crystal fragments and the use of
`this in ‘reverse crystal engineering’ has also been reported3.
`The method as executed consists of several steps.
`
`• Obtaining the single crystal data – This is not usually an issue once
`the process development stage has been reached. The ace-
`taminophen single crystal structure was obtained from
`Cambridge Structural Database (CSD). The reference code
`is HX-ACAN01 and it is the common polymorph, which
`crystallizes in the P21/c space group.
`• Simulating the powder pattern from the single crystal data – The powder
`pattern was simulated using the diffraction module in Cerius2
`(Molecular Simulations, Inc., San Diego, CA, USA) and the
`experimental patterns were subsequently ‘indexed’ from this
`reference.
`• ‘Picking’ crystals and fragments – The samples were prepared by
`picking approximately 10–25 crystals or fragments of crys-
`tals from a sample under the microscope.
`• Mounting the crystals and fragments on a powder x-ray diffraction (PXRD)
`cell – These were mounted on a PXRD sample cell (with
`adhesion required for orientation) oriented with the face(s)
`in question parallel to the cell plane.
`• Performing an appropriate PXRD scan – A PXRD scan was then run
`over the angular 2u range of interest on a Shimadzu 6000
`
`238
`
`Figure 1. Powder x-ray diffraction patterns for acetaminophen
`showing the simulated patterns as well as those from the 110, 001
`and 020 faces.
`
`diffractometer. Alternately, or in addition, an omega scan can
`be used to allow for slight miss-alignment of the crystal(s).
`The crystals are oriented on what appears to be the same
`face, a run is performed, the crystals are re-oriented, and an-
`other pattern is collected. This sequence is repeated until all
`the major faces have been scanned.
`• Comparing the data to the simulated pattern for identification of the index
`– Figure 1 shows the PXRD patterns for the oriented crystals
`and the simulated powder pattern.The Miller indices are de-
`termined from the simulated pattern. By matching the sim-
`ple patterns from the oriented crystals, the faces are identi-
`fied as the 110, 001 and 020 faces, respectively.The patterns
`can show multiple peaks at the expected angle due to height
`differences in these rather large crystals (300–500 mm).
`Also, note that there are two peaks in the 110 and 001 pat-
`terns due to the 220 and 002 planes representing the same
`family.The 020 is not preceded by an 010 peak because this
`is a systematic absence in the P21/c space group.
`• Simulated morphologies may then be used to ‘match’ the observed habit – The
`morphology predictor in Cerius2 was used to obtain an ap-
`proximate morphology for the acetaminophen crystals. The
`predictor uses the single crystal structure and a combination of
`some simple laws of crystal growth and attachment energy dif-
`ferences to simulate both the size and identity of the most
`probable faces.This was used both to aid in ‘indexing’ the faces
`and as a starting point to match the observed morphologies to
`possible simulated morphologies by achieving computational
`‘growth’ of certain faces in a preferential manner.
`
`Fractal analysis of pharmaceutical particles
`Characterization of physical properties of pharmaceutical
`materials is important because the physical properties of
`
`Merck Exhibit 2163, Page 4
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`
`
`PSTT Vol. 1, No. 6 September 1998
`
`research focus
`
`reviews
`
`individual particles may be related to
`bulk properties. For example, flowabil-
`ity is known to be affected by morpho-
`logical properties, such as particle size,
`particle shape and surface irregularity.
`Although a number of methods are
`available for characterization of particle
`size and shape, methods for the
`characterization of the surface irregu-
`larity or roughness are still poorly es-
`tablished. The authors used fractal
`analysis to characterize the surface
`roughness of pharmaceutical solid ma-
`terials. The basic theory of fractal analy-
`sis, developed by Mandelbrot in 19774,
`explains that the method is a resolution
`analysis that tracks the recurrence of
`topographical surface at different length
`scales. Traditional Euclidean geometry
`depicts a perfect straight line as a one-
`dimensional feature, an ideal plane as a
`two-dimensional feature and an ideal
`cube as a three-dimensional feature.
`Fractal dimension is a universal number
`that can be used for numerical evalua-
`tion of the degree of surface irregular-
`ity or the space-filling ability. It has
`been found that the surface and inter-
`face topographies of a large number of
`materials are fractals at the molecular
`level, and fractal analysis has become a
`widely accepted approach for the evalu-
`ation of surface roughness. In the study
`performed by the authors, the surface
`profile or topography was measured
`with an atomic force microscope
`(AFM). For several years, the AFM has
`demonstrated powerful functionality in
`many research areas, including surface
`roughness characterization. The benefit
`of using an AFM is that it is possible
`to obtain the surface profiles at the nanometer scale. It has been
`shown that the commonly used box-counting method is un-
`suitable and the power spectrum method generates relatively
`low-precision fractal dimensions for the digitized data. Thus,
`for the calculation of fractal dimension using digital data ob-
`tained form AFM, it was possible to implement the variation
`method5. The authors have used both box-counting and vari-
`ation methods and found that the latter is superior in attempts
`to calculate the fractal dimension from the AFM data.
`
`Figure 2. Three-dimensional-rendered graphics of surfaces measured with an atomic force
`microscope. (a) wet granule of caffeine and hydroxypropylmethylcellulose; (b) Ac-Di-Sol powder;
`(c) Avicel PH101 powder; (d) Di-Tab powder; (e) mannitol powder; and (f) freeze-dried mannitol
`powder. The fractal dimension values and scanning areas were: (a) 2.25 and 5 3 5 mm2; (b) 2.14 and
`2 3 2 mm2; (c) 2.10 and 2 3 2 mm2; (d) 2.13 and 10 3 10 mm2; (e) 2.15 and 20 3 20 mm2; and
`(f) 2.48 and 20 3 20 mm2.
`
`The surfaces of a range of pharmaceutical particles and gran-
`ules were examined by an AFM (NanoScope-Multi-Mode, Digi-
`tal Instruments, Inc., Santa Barbara, CA, USA) and their fractal
`dimensions were calculated. Figure 2 shows AFM images of a
`wet granule, Ac-Di-Sol powder (croscarmellose sodium, FMC,
`Newark, DE, USA), Avicel PH101 powder (microcrystalline cel-
`lulose, FMC), Di-Tab powder (dibasic calcium phosphate dihy-
`drate, Rhône-Poulenc, Cranbury, NJ, USA), and mannitol pow-
`der (Mallinckrodt Baker, Paris, KY, USA) before and after freeze
`
`239
`
`Merck Exhibit 2163, Page 5
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`
`
`reviews
`
`research focus
`
`PSTT Vol. 1, No. 6 September 1998
`
`drying. As shown in Fig. 2(a)–(e), the fractal dimensions of all
`particles and granules were in the range of 2.1 and 2.3. This
`may be due to the fact that most of the pharmaceutical particles
`and granules were placed under similar processing conditions
`and this resulted in similar levels of surface roughness. When
`mannitol particles were freeze-dried, however, the fractal di-
`mension was changed dramatically. As shown in Fig. 2(f), the
`AFM image of freeze-dried mannitol was quite different from
`that of the control mannitol. Such a difference is shown by the
`significant change in the fractal dimension. The surface of the
`freeze-dried mannitol sample was much rougher and the frac-
`tal dimension became significantly larger. In addition, the sur-
`face texture appeared to be changed by the freeze drying pro-
`cess. Figure 2(e) shows needle-shaped mannitol crystals on the
`surfaces, whereas no such mannitol crystals were observed on
`the freeze-dried mannitol surfaces [Fig. 2(f)]. These images
`demonstrated the ability of AFM to probe and obtain the three-
`dimensional surface profiles at the nanoscale. The results ob-
`tained by the authors in the fractal analysis of pharmaceutical
`samples revealed the existence of an intrinsic relationship be-
`tween the fractal dimension and the underlying processes that
`produced the material and formed the surface morphology. It
`would appear that the fractal analysis using AFM can be used to
`quantify the surface roughness and to characterize pharmaceu-
`tical materials6.
`
`Issues in bulk drug manufacture
`Once specifications and methods have been established for raw
`materials, the unit processes involved in their production can
`be examined for possible improvement or change. The current
`focus is often on understanding possible solid-state transfor-
`mations during final crystallization. Crystal engineering is an
`emerging technique that is designed to produce a narrow size
`distribution of crystals with a desired ‘shape’ or morphology.
`
`Drying of hydrates and solvates
`Drying is one of the last steps in bulk drug synthesis. Once the
`drug substance is crystallized, it is dried to produce the final
`drug substance. Drying is a unit process that can result in the
`loss of control of the manufacturing process for the drug sub-
`stance, often resulting in lot failures due to unspecified poly-
`morph and/or unsuitable particle size. Drying is particularly
`problematic if a hydrate or solvate is dried in this final step.
`Drying is even more problematic if a hydrate or solvate is dried
`to an anhydrous form in cases in which the anhydrous drug sub-
`stance exists in multiple polymorphs. In these cases different
`polymorphs can be formed due to statistical variations involving
`nucleation of different polymorphs. In order to understand more
`about the drying process, the authors have studied the drying of
`phenobarbital. Phenobarbital (5-ethyl-5-phenylbarbituric acid)
`
`240
`
`Does a phase change occur upon drying?
`X-ray diffraction
`Infrared
`Solid state NMR
`
`No
`
`No –
`check particle size
`
`Yes
`Are there different anhydrous forms?
`Polymorph screen
`Crystallize from different solvents
` – X-ray diffraction
` – Infrared
`
`Yes
`Does seeding during
`drying control the phase?
`Drying excipients
`X-ray diffraction
`
`No –
`search for other crystalline conditions
`to prepare anhydrous phase directly
`
`Yes –
`validate method
`
`Figure 3. Flow chart to guide studies aimed at controlling the drying
`process.
`
`has been reported to crystallize in as many as 13 modifi-
`cations7,8, and many of these forms were obtained by recrystal-
`lization with low levels of other barbiturates. More recent work
`suggests that there are four modifications of pure phenobarbi-
`tal that are relevant to pharmaceutical processing9.Two of these
`forms are polymorphs (Form A and B) whereas the other two
`have been identified as a monohydrate and a hemihydrate. In
`an investigation of the physiochemical stability of these forms
`at different temperature and relative humidity (RH) levels, it is
`concluded that the drying of the hydrated forms yields a mix-
`ture of polymorphs. With this background, the primary objec-
`tive of this project is to understand how a unit operation, such
`as drying, determines which solid state form is obtained and
`the factors that are instrumental in producing the desired form.
`In the first instance, the various polymorphs were prepared
`and the effects of drying at various temperatures were evalu-
`ated. For these experiments the phenobarbital monohydrate or
`hemihydrate was dried at a controlled temperature and humid-
`ity in a laboratory oven. The products were analysed by x-ray
`powder diffraction, and visual comparison of the diffrac-
`tograms was used to determine the final polymorphs present.
`The hemihydrate changes to a mixture of Forms A and B after
`storage for three hours at 508C and 0% RH. The appearance of
`the mixture remains constant when drying is extended to up to
`15 days, and only when the hemihydrate is heated to 1008C is it
`possible to observe a transformation to Form B.The behavior of
`the monohydrate upon storage at 508C and 0% RH is essentially
`
`Merck Exhibit 2163, Page 6
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`
`
`PSTT Vol. 1, No. 6 September 1998
`
`research focus
`
`reviews
`
`the same as the hemihydrate. A mixture of forms A and B is
`observed after several hours and this composition is un-
`changed after 30 days of storage. Similar results occur whether
`the monohydrate was prepared by the wet milling of either
`Form A or B.
`In summary, it is clear that drying of either of these hydrates
`results in mixtures of crystal forms. Consultation of the flow
`charts in the International Conference on Harmonization Q6A
`Draft Guidance on Specifications establishes that such a process
`is out of control and would be problematic for the production
`of bulk drug substance.Additional studies are in progress to find
`ways to control the drying process and to control the crystal
`form. Figure 3 contains a flow chart, based on this research, to
`offer guidance in the drying process. This flow chart should be
`used to help avoid instances in which drying is an uncontrolled
`step in the production of drug substances or drug products.
`
`Processing and process control
`Control of processes has been both a concern and a strong
`point within the pharmaceutical industry. Pharmaceutical sci-
`entists operate under some of the strictest guidelines in exist-
`ence and have done well in attempts to develop expertise to
`achieve control of the processes in current use.The approaches
`for the future have the general character of trying to transfer
`the expertise, in part, to monitoring systems. This may take
`much of the subjectivity out of the determination of process-
`ing control and makes it easier for regulatory agencies to verify
`the validation activities with less review. It may also require
`less specific experience by the operator to execute the process
`and thus help to reduce dependence on individuals for pro-
`duction. The acceptance of new processing techniques is
`slower because of both the high degree of reliability that must
`be demonstrated and the associated costs.Two examples of new
`technologies in development are roller compaction and super
`critical fluid (SCF) particle size reduction. Roller compaction is
`being used increasingly in production sites, whereas SCF tech-
`niques are only now becoming a practical option (for exam-
`ple, they are in use at Bradford Particle Design in the UK and in
`the work presented here).Whatever the control and processing
`issues, the two are inexorably linked as more sophisticated pro-
`cesses require more advanced monitoring in order to maintain
`control. Linking all of these advances with computer inte-
`gration suggests the possibility of achieving parametric release
`of product: this means less analysis cost, resources and time, as
`well as full tracking of the physico-chemical and equipment
`parameters throughout the production process.
`
`Near-infrared determination of water in formulations
`The determination of moisture in materials was the first appli-
`cation of near-infrared (NIR) spectroscopy and this remains its
`
`most important use. Because of the improvement in NIR in-
`strumentation and the ongoing research in this field, this
`method is now increasingly applied in the analysis of a range
`of pharmaceutical systems and problems, and NIR can be a
`useful technique in process and quality control10–18. NIR meas-
`ures the overtones and combinations of the vibrational modes
`of a molecule, principally those involving bonds with hydro-
`gen (i.e. C–H, N–H and O–H). Because of the low absorptivi-
`ties of these bands, in contrast to mid-infrared measurements
`samples need not be diluted. Furthermore, NIR radiation is not
`absorbed by ordinary glass and is readily scattered by particles.
`Thus diffuse-reflectance spectroscopy is the most prominent
`sampling technique for solids and it is therefore well suited for
`on-line process monitoring. Finally, NIR spectroscopy may be
`a low cost method because it is non-destructive, fast and
`requires no special sample preparation.
`The absorption spectrum of pure water shows, in the main,
`two strong absorption bands in the NIR region of ~750–2500 nm
`(4000–13,330 cm21), one band at 1940 nm (5150 cm21),
`and one band at 1450 nm (6900 cm21). The peak position is
`temperature dependent and is shifted to higher frequencies
`when water associates with other molecules by hydrogen
`bonding. Quantification of water is based on the measurement
`of these bands and other reference values of the spectrum
`while applying a proper mathematical algorithm to minimize
`variabilities in the reflectances caused by, for example, particle-
`size effects.
`Many commercial NIR-sensors, such as the MM55 (Infrared
`Engineering, Inc., Irwindale, CA, USA), have been designed for
`the on-line monitoring of water in materials19. These instru-
`ments are fast in response (real-time monitoring) and, because
`they combine signal processing and readout, they are compact
`and easy to handle. However, successful implementation in a
`continuous on-