Meeting Report Dissolution Testing in Drug Product ... · product release and stability testing are highlighted in the ﬁrst two summaries of the Day 1 podium presentations. Additionally,

Meeting ReportTheme: Dissolution and Translational Modeling Strategies Enabling Patient-Centric Product DevelopmentGuest Editors: Marilyn N. Martinez, Sandra Suarez, and Andreas Abend

Dissolution Testing in Drug Product Development: Workshop Summary Report

Andreas Abend,1,14 David Curran,2 Jesse Kuiper,1 Xujin Lu,3 Hanlin Li,4 Andre Hermans,1 Pramod Kotwal,1

Dorys A. Diaz,5 Michael J. Cohen,5 Limin Zhang,3 Erika Stippler,6 German Drazer,7 Yiqing Lin,8

Kimberly Raines,9 Lawrence Yu,9 Carrie A. Coutant,10 Haiyan Grady,11 Johannes Krämer,12

Sarah Pope-Miksinski,13 and Sandra Suarez-Sharp9

Received 22 October 2018; accepted 18 December 2018

Abstract. This publication summarizes the proceedings and key outcomes of the first day(BDay 1^) of the 3-day workshop on BDissolution and Translational Modeling StrategiesEnabling Patient-Centric Product Development.^ The overall aims of the workshop were tofoster a productive dialog between industry and regulatory agencies and to discuss currentstrategies toward the development and implementation of clinically relevant dissolutionspecifications as an integral part of enhanced drug product understanding and effective drugproduct life-cycle management. The Day 1 podium presentations covered existing challengesand concerns for implementing highly valuable, yet often unique and novel experimentaldissolution setups as quality control tools. In addition, several podium presentationshighlighted opportunities to replace conventional dissolution testing with surrogate testmethods to enable robust drug product and process understanding within the context ofquality by design (QbD), new manufacturing technologies, and real-time release testing(RTRT). The topics covered on Day 1 laid the foundation for subsequent discussions whichfocused on the challenges related to establishing an in vitro–in vivo link and approaches forestablishing clinically relevant drug product specifications which are becoming an expectationin regulatory submissions. Clarification of dissolution-related terminology used inconsistentlyamong the scientific community, and the purpose of various testing approaches were keydiscussion topics of the Day 1 breakout sessions. The outcome of these discussions along withcreative ways to overcome challenges related to bridging Bexploratory dissolutionapproaches^ with methods suitable for end-product control testing are captured within thisreport.

KEY WORDS: biorelevant; biopredictive; clinically relevant dissolution; discriminating power; surrogatefor dissolution.

Guest Editors: Marilyn N. Martinez, Sandra Suarez, and AndreasAbend

Electronic supplementary material The online version of this article(https://doi.org/10.1208/s12248-018-0288-4) contains supplementarymaterial, which is available to authorized users.

1 Pharmaceutical Sciences, Merck & Co., Inc., West Point, Pennsyl-vania 19486, USA.

2Analytical Sciences and Development, GlaxoSmithKline, King ofPrussia, Pennsylvania 19406, USA.

3 Drug Product Science and Technology, Bristol Myers SquibbCompany, New Brunswick, New Jersey 08903, USA.

4 Vertex Pharmaceuticals Inc., 50 Northern Ave., Boston, Massachu-setts 02210, USA.

5Worldwide Research and Development, Global Chemistry andManufacturing Controls, Pfizer Inc., Eastern Point Road, Groton,Connecticut 06340, USA.

6United States Pharmacopeia, 12601 Twinbrook Parkway, Rockville,Maryland 20852, USA.

7Department of Mechanical and Aerospace Engineering, Rutgers,The State University of New Jersey, New Brunswick, New Jersey,USA.

8Analytical Development, Biogen Inc., Cambridge, Massachusetts02142, USA.

9Office of Pharmaceutical Quality, Center for Drug Evaluation andResearch, Food and Drug Administration, Silver Spring, Maryland20993, USA.

10 Small Molecule Design & Development, Eli Lilly & Co., LillyCorporate Center, Indianapolis, Indiana 46285, USA.

11 Takeda Pharmaceutical International, 35 Lansdowne Street, Cam-bridge, Massachusetts 02139, USA.

12 Eurofins-PHAST GmbH, Kardinal-Wendel-Str. 16, 66424, Hom-burg, Saarland, Germany.

13 AstraZeneca Pharmaceuticals, One MedImmune Way, Gaithers-burg, Maryland 20878, USA.

14 To whom correspondence should be addressed. (e–mail:[email protected])

The AAPS Journal (2019) 21:21 DOI: 10.1208/s12248-018-0288-4

1550-7416/19/0000-0001/0 # 2019 American Association of Pharmaceutical Scientists

http://dx.doi.org/10.1208/s12248-018-0288-4

http://crossmark.crossref.org/dialog/?doi=10.1208/s12248-018-0288-4&domain=pdf

INTRODUCTION

In May 2017, a 3-day workshop titled BDissolution andTranslational Modeling Strategies Enabling Patient-Centric DrugProduct Development^ convened at the University of Maryland’sSchool of Pharmacy inBaltimore,MD.Themeetingwas led by theCenter for Excellence in Regulatory Sciences and Innovation (M-CERSI) and co-organized by members of the United States FoodandDrugAdministration (USFDA),EuropeanMedicinalAgency(EMA), and the International Consortium for Innovation andQuality in Pharmaceutical Development (IQ). A high-levelworkshop summary describing the overall challenges of patient-centric drug product development and strategies for linking in vitrotesting and in vivo performance was recently published (1). Inshort, this 3-day meeting provided a discussion forum for scientistsworking in the industry, academia, and regulatory agencies todebate and advance the understanding of the following criticaltopics:

1. The Role of Dissolution Testing Throughout DrugProduct Development (Day 1)

2. The Need for Establishing the In Vitro–In Vivo Link(Day 2)

3. Regulatory Applications of Clinically Relevant Disso-lution Testing (Day 3)

The agenda for all 3 days included a balanced mix ofpresentations from the industry, academia, and regulatorshighlighting the utility of in vitro dissolution testing in productdevelopment as well as concerns related to its use for productperformance understanding. Most importantly, the breakoutsessions provided a forum for participants to elaborate,rationalize, and challenge current industry practices as wellas regulatory expectations. These breakout sessions weredesigned to provoke new ideas related to in vivo productperformance understanding and to develop meaningful drugproduct specification approaches in the context of current and

emerging new technologies as well as regulatory trends. Theoutcomes of Day 2 and Day 3 discussions are the subject ofseparate publications (2,3).

In the opening presentation, Dr. Lawrence X. Yu, fromthe US FDA, briefly elaborated about the history ofdissolution testing and the need for patient-centric dissolutiontesting in product development. Since its development in thelate 1950s and early 1960s, and its acceptance by the UnitedStates Pharmacopeia (USP) Convention in 1970 (4,5), in vitrodissolution testing has been used as a quality control measurefor solid oral dosage forms. In recent times, it has beenevolving as an invaluable tool to forecast in vivo performanceof drug products and to ascertain the need for in vivocomparative bioavailability and bioequivalence studies. Ac-cording to Dr. Yu, dissolution can link product quality toin vivo performance through in vitro–in vivo correlations/relationships (IVIVC/IVIVR) developed either via conven-tional paths or using physiologically based pharmacokineticabsorption modeling (PBAM).1 Therefore, it is an essentialtest to realize quality by design for solid oral dosage forms(6,7). Recent developments in the simultaneous measurementof in vivo solubility/dissolution in the gastrointestinal tractand drug concentration in plasma provide opportunities tobetter design in vitro dissolution methods and development ofimproved mechanistic oral drug absorption models (8). Dr.Yu remarked that the FDA has recently granted biowaiversand standardized the dissolution testing requirements forhighly soluble drugs designed as immediately release dosageforms (9). However, for poorly soluble drugs designed asimmediate release or modified release products—althoughIVIVC is possible—in practice, it is difficult to achieve.Therefore, in Dr. Yu’s opinion, future drug product develop-ment efforts should focus on the measurement of in vivosolubility and dissolution as input toward the development ofin vitro dissolution methods and mechanistic absorptionmodels.

Dr. Sarah Pope Miksinski2 discussed the overall contextfor clinical relevance (10) by tying the five strategic prioritiesof the Office of New Drug Products to the broader prioritiesof the Office of Pharmaceutical Quality (11). In addition, shereinforced the importance of the Blink to the patient,^ both interms of various ongoing initiatives as well as regulatorydecision-making and general operations. Her presentationfocused on the fundamental link to patient expectations forquality, which include a product being safe and effective,exhibiting performance as labeled, maintaining expectedperformance throughout the shelf life, being manufacturedin a manner that assures quality, and being available asneeded. Dr. Pope-Miksinski also outlined the essentialconcepts of the current dialog on clinical relevance, includingdata requirements and recommendations, robust and trans-parent risk communication, the presence/acknowledgementof uncertainty, the efficiency of interactions, and payingappropriate attention to Bthe big picture.^ She discussed the

Abbreviations API, Active pharmaceutical ingredient; AUC, Areaunder curve; BA, Bioavailability; BCS, Biopharmaceutical Classifi-cation System; BE, Bioequivalence; Cmax, Maximum concentration;cGMP, Current good manufacturing practices; CMA, Critical mate-rials attribute; CPP, Critical process parameter; CQA, Critical qualityattribute; CRDPS, Clinically relevant drug products specification(s);CSOP, Engineering Research Center for Structured Organic Partic-ulate Systems at the New Jersey State University at Rutgers, NJ;EMA, European Medicines Agency; ER, Extended release; FDA,Food and Drug Administration; ICH, International Council forHarmonization of Technical Requirements for Pharmaceuticals forHuman Use; IQ, International Consortium for Innovation & Qualityin Pharmaceutical Development; IR, Immediate release; IVIVC,In vitro–in vivo correlation; IVIVR, In vitro–in vivo relationship; M-CERSI, Maryland-Center for Excellence in Regulatory Sciences andInnovation; M&S, Modeling and simulation; NIR, Near infrared;SUPAC, Scale-up and postapproval changes; PAT, Process analyticaltechniques; PBPK model, Physiologically based pharmacokineticmodel; PBAM, Physiologically based absorption model; PCA,Principal component analysis; PK, Pharmacokinetic; PLS, Partialleast square; PSD, Particle size distribution; QC, Quality control;QbD, Quality by design; QRA, Quality risk assessment; RTRT, Real-time release testing; RTQA, Real-time quality analysis; TPP, Targetproduct profile; USP, United States Pharmacopeia

1 The term PBAM applied to drug product quality isevolving. The terms PBAM and physiologically basedbiopharmaceutics modeling (PBBM) were used inter-changeably in the subsequent manuscripts following theworkshop and are part of the theme.

2 At the time of the workshop, a director at the US FDA

21 Page 2 of 12 The AAPS Journal (2019) 21:21

penultimate benefit–risk framework, which balances potentialrisks to quality with availability to patients/consumers. Dr.Pope-Miksinski concluded her presentation by outlining fivekey concepts of clinical relevance, each relating to context,connections, and collaboration in the regulatory landscape.

Companies represented by IQ are, in principle, alignedwith the position presented by Dr. Yu that dissolution testingplays a critical role in enhanced product understanding (e.g.,QbD) (12). Dissolution testing under physiologically relevantconditions is common practice in many companies especiallyin early development. When such methods are applied, a linkbetween in vitro dissolution testing and in vivo performancecan be achieved in many cases. However, expectations forimplementing these methods for routine quality controlpurposes are concerning to the pharmaceutical industry, asthese methods may lack the robustness expected for QCmethods (13). In addition, there is general concern overglobal regulatory acceptance due to the fact that thesemethods are frequently noncompendial in nature. Currentindustry practices using biorelevant dissolution in early stagesof development and strategies toward bridging these methodswith globally acceptable dissolution specifications for routineproduct release and stability testing are highlighted in the firsttwo summaries of the Day 1 podium presentations.

Additionally, there is an emerging trend in the industry toexplore alternatives to dissolution testing and to apply themduring product development to ensure product quality insteadof relying on traditional dissolution testing. This goal can beachieved by focusing on a combination of in-process and/or at-line analytical tests and in silicomodeling resulting in predictivedissolution models. In many instances, it can be demonstratedthat these tests and/or models are as predictive of in vivoperformance as traditional or nontraditional dissolutionmethods. Hence, they support establishing both clinicallyrelevant drug product specifications and robust manufacturingcontrol strategies. There is currently limited experience both inthe industry and within regulatory agencies to use these testsinstead of traditional dissolution. However, such alternativeapproaches, underpinned by a deep understanding of theinterrelationship between in vitro dissolution testing, otherquality attributes/in-process controls, and in vivo drug productperformance assessment, are also expected to be enablers ofcontinuous manufacturing and RTRT. Compared to the otherCQAs required to empower RTRT (i.e., content uniformity,purity, etc.), modeling dissolution performance may be the mostchallenging, as tablet dissolution is often influenced by severalmaterial attributes and process parameters. Examples of the useof surrogates for dissolution testing as well as predictivedissolution modeling are described in the sections followingthe summaries about dissolution testing under physiologicallyrelevant conditions.

SUMMARY OF PODIUM PRESENTATIONS

The Role of Dissolution Testing in Early FormulationScreening (14)

The principal driver behind conducting in vitro dissolu-tion studies in the early phase of product development is topredict in vivo performance of formulation candidatesentering phase I clinical trials. With the assumption that

in vivo dissolution is meaningful for PK performance inhumans, the minimum expectations for early stage dissolutionmethodology are to identify what aspects of the drugsubstance, formulation composition, and process are mostimportant to achieve the desired in vivo performance of thedrug product. The in vitro dissolution studies may comple-ment or substitute preclinical (animal) models, and they areperformed to guide formulation development and optimiza-tion and to select the best candidates for BA studies.

When it is known that in vivo dissolution is the rate-limiting step in the absorption process, one approach forformulation candidate screening is to conduct dissolution atthe solubility limit of the drug substance (15). This B1Xdissolution^ (16) means that the target drug concentration ofthe in vitro dissolution experiment is equal to the solubilitylimit of the API. Since the dissolution rate is meaningful forin vivo performance, formulation differences that make ameasurable impact on the rate at which the drug reaches itssolubility limit are explored.

In the example shown in Fig. 1, 1X biorelevant dissolutionwas conducted using prototype formulations where in vivoperformance was known to be dependent on the dissolution rateof the API. Spray-dried intermediates with varied formulationattributes were preparedwith drugAand a polymer (copovidone),both with and without sodium lauryl sulfate (SLS) in theamorphous solid dispersion. These intermediates were thenprocessed into tablets (FM 1 tablet and FM 2 tablet). In addition,one tablet (FM 3 tablet) and one capsule formulation (FM 3capsule) containing only amorphous API with traditional excipi-ents were prepared, in order to investigate if dosage formscontaining amorphous API give comparable in vivo exposures toformulations containing the amorphous API in a spray-driedintermediate. Dissolution testing was performed using a two-stageB1X^ dissolution approach on the three tablet formulations. Thedissolution rate of the formulations in the second stage (FaSSIF) ofthe two-stage biorelevant dissolution experiment is dependentupon the ability of the formulation to form amorphous particles inthe first stage (SGF) of the test. The dissolution results clearlyrevealed a difference in these formulations’ ability to deliveramorphous particles in the SGF stage and therefore differentiateddissolution rate in the second stage. The observed rank–orderrelationship between dissolution rate demonstrated a clear corre-lation with systemic exposure, further supporting that for this drug,dissolution is critical for drug product in vivo performance (16).

Two multicompartment models were also discussed inthis presentation. In the first example, a transfer modelsystem was established to investigate the in vivo behavior ofweakly basic compounds. Preliminary data showed promisingresults to support a transfer model as an alternative way toestimate in vivo precipitation in the intestinal compartmentfor these compounds (Fig. 2) (17). This model accuratelysimulates the gastric retention time and transfer rate of drugsinto the intestinal tract. The transfer rate is especiallyimportant for compounds that may precipitate in the stomachdue to possible pH changes of the gastric fluid. A couple ofopportunities for this methodology were described. Forexample, the transfer model enabled the development of anin silico model, including a full mathematical model todescribe simultaneous transfer/precipitation process as wellas methodology to describe formulations and drugs usingenabling formulation strategies.

21 Page 3 of 12The AAPS Journal (2019) 21:21

The second multicompartment model discussed was anartificial stomach duodenum (ASD) in vitro system thatmimics the dynamic conditions of the human gastrointestinaltract, as well as the biorelevant fluids present in these systems(18). In the schematic shown in Fig. 3, the green circlesrepresent pumps that introduce biorelevant fluids at eachstage. The drug concentration is measured in the stomach andduodenum compartment via UV fiber optic probes. The ASDconcept captures supersaturation, precipitation, and dissolu-tion phenomena as they occur in vitro and relates them tohow they may occur in vivo. Several enhancements for theASD methodology were suggested, including standardizationof fluid compositions, solids transport, agitation rate, and fedvs. fasted simulations. Also discussed were enhancements tothe physical system including additional compartments, theability to adjust transit time, automated low-volume sampling,and methods to simulate removal of aqueous drug from thesystem (absorption).

Dissolution Methodologies from Biorelevant to QualityControl (19)

As discussed earlier, biorelevant dissolution plays a keyrole in early development for formulation screening,biopharmaceutics risk assessment, and formulation develop-ment toward achieving the targeted product profile (TPP)(20). In addition to the properties of the drug substance andthe drug product, both biorelevant media and hydrodynamicsneed to be considered when applying these dissolutionmethods. As a result, the experimental conditions one mayuse range from simple (i.e., SGF and using USP apparatus Ior II for tablets containing highly soluble drug substances) tohighly sophisticated setups (for example TIM-TNO (21)) withdifferent biorelevant media for tablets formulated with poorlysoluble drug substances.

On the other hand, dissolution is also a key test toconfirm batch-to-batch consistency and drug product qualityat release and throughout its shelf life. Companies thereforeimplement dissolution specifications following current

regulatory guidance in routine quality control laboratoriesusing standard equipment and under conditions described inapplicable compendia.

The coexistence of several sets of dissolution tests—a setof internal tests to enable product development and anBofficial^ test to ensure product quality control—begs thequestions of the degree of alignment between these sets, andwhich set represents the best description of the dissolutionbehavior of the product. To answer these, and perhapsadditional related questions, close examination of bothcurrent industry practice with regard to dissolution testingapproaches and their significance toward in vivo drugperformance understanding is necessary.

A comparison of key attributes for the two sets ofmethods is depicted in Table I.

Traditionally, GMP-compliant QC methods are per-formed in compendial dissolution devices and simple buff-ered media, with a regulatory expectation for Bsinkconditions^ and complete drug substance release within ameaningful timeframe. In contrast, biorelevant dissolutionmethods are often performed in Bnoncompendial^ equip-ment using complex biorelevant media. These methods areBuniversal^ (i.e., nonproduct specific) and companies imple-ment them as part of their formulation developmenttechnology platform to assess in vivo performance andbioperformance risks. Multiple biorelevant methods may bepart of the platform technology and thus may be applied tounderstand in vivo performance of a given formulationconsidering, for example, drug–drug interactions or foodeffects. The in vitro release profiles may be monitored atnonsink conditions and allow formulation ranking even incases where full release is not achieved (see for exampleFig. 1).

Typical QC and biorelevant methods are not onlydifferent with respect to their purpose and experimentalconditions under which they are performed, but the require-ments and anticipated variabilities in individual data pointsmay be significantly different, as discussed above. In partic-ular, data generated under biorelevant method conditions

Fig. 1. B1X^ dissolution of tablets and capsules used in clinical studies in biorelevant media and theirassociated PK data


often reveal a high degree of variability (13), and therefore,the methods may not be sufficiently robust for transfer to aroutine QC lab. Additionally, since the equipment and thereagents are not standardized, implementing Bsystemsuitability^ or other meaningful performance criteria maynot be practical. Nevertheless, in some cases, a biorelevantmethod may be applied as a routine QC test upon consider-ation of a few questions, including regulatory expectation forqualifying and justifying noncompendial equipment, settingacceptance criteria in the absence of full drug release, theappropriate use of the standard staged criteria (e.g., Q+5 forStage 1), and the potential for a biorelevant method withhigher variability to meet nonstandard analytical validationcriteria.

Efforts continue to be made by both industry and theregulatory authorities to build a bridge across the significantgaps from the biorelevant method(s) to a QC method thatcan detect and control for changes in CPPs and CMAsthereby ensuring acceptable in vivo performance (clinicallyrelevant) for BCS 2 and BCS 4 drugs and modified releaseformulations. In order to close the gap between biorelevantand QC methods, several approaches have been exploredincluding (a) proactive collaboration between functionalareas involved in pharmaceutical product development, (b)

phase-appropriate Bfit-for-purpose^ dissolution methodol-ogy, (c) application of QbD principles during development,and (d) systematic transition or evolution from biorelevantdissolution to QC dissolution.

When the goal is to work toward the development ofdissolution methods that are clinically relevant, a decisiontree may help in the development process (see example inFig. 4). The process starts with the selection of biorelevantmethods to evaluate prototype formulations, and thennarrows them down to one method that can detect the keyrisks related to in vivo performance. The evolution of thebiorelevant method(s) to the QC method is largely drivenby systematic simplification of the testing procedure whilemaintaining or building discrimination for the productattributes that can be measured by in vitro dissolution andare most likely to impact PK.

The Use of Surrogates for Dissolution Testing for ImmediateRelease Formulations—When Is It Feasible? (22)

When investigating the sensitivity of dissolutionmethods toward changes in CPPs and CMAs, companiesfrequently use an array of analytical tools in an attempt tocorrelate the data to the rate and extent of in vitro drug

Fig. 2. Example of a simple dissolution transfer model to study the in vivo behavior ofdrug products containing weakly basic compounds

Fig. 3. Schematic of the artificial stomach duodenum (ASD) system


release. In case these measurements demonstrate a corre-lation to drug product dissolution, companies may imple-ment these analytical methods as part of manufacturingprocess controls, or finished product testing. Some of thesetechniques as well as the specific steps which are a subsetin the overall in vitro dissolution event and the potentialCPPs and CMAs they are probing are shown in Fig. 5.Since these measurements are typically readily available,sensitive, accurate, precise, robust, and rapid, it is highlydesirable from an industry perspective to apply theminstead of traditional dissolution while maintaining thesame level of quality assurance. Examples of the use ofthese novel analytical tools as surrogates for dissolutiontesting either by themselves or as part of a predictivedissolution model are discussed in the following section.

From a regulatory standpoint, currently only BCS class 1and 3 compounds allow for surrogate testing in the form ofdisintegration as defined by the ICH Q6 decision tree (23).The main rationale for limiting surrogate tests to highlysoluble compounds stems from the fact that overall tabletdissolution is largely independent of the rate of APIsolubilization and the overall dissolution profile is dictatedby tablet disintegration. Hence, in many cases, a dissolution–disintegration relationship can be established. It has beenshown that for certain IR tablets, disintegration is moresensitive toward process factors such as hardness, as com-pared to dissolution. In case a clear correlation for hardnessand disintegration can be established, tablet hardness canreplace disintegration as part of RTRT, which has beenaccepted for several products, albeit only in some markets.

Table I. Purpose and Key Attributes of Biorelevant and QC Dissolution Methods

Biorelevant dissolution Quality control dissolution

Purpose Predicting in vivo performance Ensuring batch-to-batch consistencyDevice Compendial and noncompendial CompendialMedium Biorelevant media Conventional buffers w/ or w/o surfactantMethod development Conditions chosen to mimic the environment

of the GI tractConditions chosen to detect process and stabilitychanges specific for a given product

Profile Drug substance amount released varies undernonsink condition; used to rank order

Full release expected within a defined amount of time;usually at 3–5 sink

Early phase Formulation selection, CMA, CPP,and CQA identification

Clinical batch release as required by specification

Later phase Potential to establish IVIVC/IVIVR; guide designof BA or BE studies if necessary

Clinical batch release; justification of commercialdrug product specifications and if possible, use toestablish IVIVC/IVIVR

Refer to the breakout summary for the participants’/authors’ assessment on the differences among several dissolution terminologies

Fig. 4. Flowchart for the systematic transition from biorelevant dissolution methods to a single QCdissolution method


However, for BCS class 2 and 4 products, the APIdissolution rate is assumed to be a major factor in the overalltablet dissolution event. Thus, disintegration might not be theoverall rate-limiting step for low-solubility compounds andcannot be used to ensure consistent release of the API fromthe drug product. Regardless of the solubility of theAPI—and as a general approach—it was proposed to usemechanistic dissolution understanding of the drug product asa guiding principle to determine the dissolution rate-limitingstep and to select proper surrogate characterization tests.Figure 5 depicts the overall dissolution mechanism ofimmediate release solid oral dosage forms as well as potentialsurrogate tests for each dissolution step along with examplesof raw materials and drug product material attributesimpacting these steps.

To demonstrate the potential application of surrogatetesting approaches for BCS class 2 and 4 compounds, casestudies where enhanced product understanding justifies theuse of disintegration or CMAs/or CPPs as quality predictorsfor dissolution testing were presented. The presented exam-ples included applications from the industry and academia onusing first principle modeling as well as empirical methodssuch as fitting of experimental data to a Weibull function ormultivariate approaches for dissolution modeling of dosageforms containing low-solubility drugs. It was demonstratedthat first principle modeling based on mass transfer modelsutilizing commercially available software (DDD Plus™) canresult in good predictability of dissolution performance invarious surfactant-containing media. This approach can beespecially useful during dissolution method developmentwhere simulated dissolution profiles could replace actual

dissolution experiments to predict in vitro dissolution perfor-mance of the drug under study.

Two case studies where disintegration was a suitablesurrogate for dissolution testing for low-solubility compoundswhen formulated as amorphous solid dispersion formulationswere presented to demonstrate that a surrogate method canbe more sensitive than dissolution testing and is appropriatefor enhanced product understanding as well as assurance ofproduct quality. In these examples, it was demonstrated thatthe tablet disintegration rate was rate limiting and dictatedthe overall dissolution profile. In addition to surrogateapproaches, case studies on using first principles such as amodified Noyes–Whitney dissolution model and empiricaldissolution models such as a modified Weibull to predictdissolution performance of drug products with fast- and slow-release profiles were presented and discussed. The develop-ment of a dissolution model requires extensive efforts tocorrelate the CMA and CPP with model parameters whichallows for the development of an enhanced product controlstrategy justifying the elimination of dissolution testing forproduct release.

Enabling Real-Time Quality Assurance with NIR-BasedPrediction of Dissolution for Tablets Made by ContinuousDirect Compression (24)

The development of predictive dissolution models toenable RTRT (sometimes referred to as Breal-time qualityassurance^ or RTQA) of pharmaceutical solid dosage formsis essential in continuous manufacturing systems to enableclosed-loop process control.

Fig. 5. Schematic of the dissolution of an IR tablet containing granulated API and associated rate constants: tabletdisintegration into granules, dissolution (disintegration) of granules, and release of API particles followed by APIsolubilization


A general method to develop formulation-dependentstatistical models that rely on nondestructive spectral mea-surements as well as process parameters to predict thedissolution profile of individual pharmaceutical tablets wasestablished at the Engineering Research Center for Struc-tured Organic Particulate Systems (C-SOPS). The step-by-step method is summarized below for an immediate releaseformulation of a highly soluble API; however, it was notedthat an analogous procedure was successfully followed for anER formulation.

Stage 1. Design of experiments. First, it is necessary todetermine the target processing conditions ofthe solid tablets, including formulation compo-sition. For example, tablets composed of 9%acetaminophen, 90% lactose, and 1% magne-sium stearate by weight were investigated. Thetarget compression force was 24 kN for 350 mgtablets. Other processing variables included theblender speed (200 rpm) and the feed framespeed (25 rpm). A more detailed description ofthe continuous manufacturing process in thiscase can be found in Pawar et al. (25). Then, theinitial step consists of designing the necessaryexperiments to explore the effect that variationsin formulation and/or processing conditionswould have on dissolutions performance. It iscrucial to identify the most important variablesthat could result in changes to the dissolutionprofile. In the case discussed, several factorsincluding API concentration, compaction force,blender speed, and the rotation speed of thefeed frame were varied. In each case, threedifferent levels for each variable were explored.

Stage 2. Dissolution and spectral measurements. Thetablets manufactured under the conditions de-termined in step 1 are then scanned in trans-mission mode using NIR spectroscopy. It wasnoted that alternative spectral methods, such asRaman spectroscopy, could be used dependingon the formulation and tableting process underconsideration. It is also possible, depending onformulation, to use reflectance NIR from bothsides of each tablet. Following the spectroscopystudies, dissolution profiles for all the tabletswere obtained using standard compendial meth-odologies described in the USP.

Stage 3. Model development. First, each dissolutionprofile is fitted to an ad hoc dissolution model.In the case discussed here, a Weibull model withtwo parameters provided a good fit to theexperimental data. The application of an alter-native model independent approach is alsopossible (see Pawar et al. (25)—for a modelindependent approach based on the level andshape of the dissolution curves). At the sametime, principal component analysis was per-formed on the spectral data obtained from thetablets. Finally, a multilinear regression wasperformed between the principal componentsidentified from the spectral data (retaining the

first 3 components) and the fitting parameters ofthe dissolution curves. This provided a statisti-cal model to predict dissolution profiles forindividual tablets.

Stage 4. Validation. The model was finally used topredict the dissolution profile of six indepen-dent tablet variants manufactured at the targetconditions. In all cases, good agreement wasobtained between the predicted and actualdissolution profiles. The profiles were thencompared using the standard f1 and f2 factorswith excellent results. The measured f2 valuesfor individual tablets ranged from 75 to 79 andthe f1 values ranged from 3 to 10.

The methodology described here can be utilized forother formulations and the C-SOPS is actively pursuing othercase studies, including ER tablets. A remarkable advantageof the proposed approach is its nondestructive nature, whichwould enable investigating and correlating other criticalproperties with dissolution as well as testing for dissolutionperformance at multiple times during the manufacture andproduct shelf life. Note that the same methodology could beapplied to batch processing.

Dissolution Modeling in Support of Real-Time ReleaseTesting for a Fixed-Dose Combination Product (26)

In this case study, dissolution modeling of a fixed-dosecombination (FDC) tablet with two APIs was presented.Although the FDC tablets were made through a co-granulation process, two separate offline dissolution methodswere applied for the two APIs, respectively. Correspondingly,two RTRT dissolution models have been developed. A com-prehensive understanding of the drug product formulation andmanufacturing process was essential to establish the RTRTmodel. As depicted in the fishbone diagram below (Fig. 6), theinputs of factors for dissolution modeling outlined in the boxesare outputs of many process parameters.

For each input factor that could potentially influencetablet dissolution, process analytical techniques (PAT) havebeen implemented, with measurements taken at differentstages of the process, as shown in Table II.

Similar to the presentation earlier (24), a stepwiseapproach was also taken in developing the dissolution model.Tablets made from a DOE run, where the input factors arepurposefully varied, were first measured by the offlinedissolution method, i.e., USP dissolution. From the referencedissolution profiles collected, dissolution rates (Z) weredetermined by fitting through a modified Noyes–Whitneyequation, as shown below:

Z ¼ dLCdt

¼ z p−LCð ÞnðS−DoseVol

LCÞ ð1Þ

LC = % dissolved at time t, Dose = target dose, Vol =volume of the dissolution medium (fixed parameter), S =solubility of the API (fixed parameter), n = shape factor(fixed parameter), z = rate factor (fitted parameter), and p =plateau (fitted parameter).


Step 2 of the model development was to take themeasured attribute data by PAT (as in Table II) and build achemometrics (partial least square or BPLS^) model topredict the dissolution rate (Z). The goodness of fit wasmeasured by comparing the PLS model predicted dissolutionrate and the reference dissolution rate determined in step 1.With the predicted dissolution rate and the modified Noyes–Whitney equation, the predicted dissolution profiles werereconstructed as shown for example in Fig. 7.

The stepwise approach shown above was included inrecent market applications, and the proposed incorporationof the predictive dissolution model as part of the product’sRTRT has obtained regulatory approval in several markets.

SUMMARY OF BREAKOUT SESSIONS

A comprehensive list of potential topics and questionswas collected prior to the meeting from IQ membercompanies, the FDA and EMA. These were prioritized andsplit into two groups with the intent to allow adequate timefor discussions. The topics selected for the Day 1 breakoutsessions represented a balanced mix of questions fromindustry and regulatory agencies, focusing mainly on termi-nology and challenges related to bridging biorelevant and QCmethods. Note that the focus of these discussions was to gainan understanding of best practices to define the experimentalconditions under which the dissolution methods are used.Nevertheless, the definitions of the various dissolutionmethods and their applicability in product development aresimilar to the definitions outlined in the recent high-levelsummary of this workshop (1).

Breakout session subgroups A1 and A2 concluded theirdiscussions with proposed definitions for commonly useddissolution-related terminology:

1. Biorelevant dissolution method(s): a set of experi-mental in vitro method conditions (media and equip-ment) that mimic the physiological environment thedrug encounters upon ingestion. The participants’opinion was to avoid the term biorelevant but rathercall these physiologically based dissolution (test)methods.

2. Discriminating dissolution method: an in vitro disso-lution method that can detect variations in CPPs andCMAs that potentially have an impact on in vivoperformance. This method is used to establish qualitycontrol dissolution specifications.

3. Quality control method: method conditions that arebased on a discriminating dissolution method. Toavoid confusion of the term Bquality control^ fordissolution methods used in product development, itwas decided to avoid this term altogether. The qualitycontrol aspect of the dissolution specification ismeaningful in a commercial environment but seemsto be of little value during development. The term QCmethod should be replaced therefore with regulatoryapproved method.

4. Clinically relevant dissolution method: a set of exper-imental in vitro dissolution conditions which aresensitive toward changes in CPPs and CMAs with ademonstrated link to in vivo drug product perfor-mance (e.g., Cmax, AUC). This link can be based onmodeling (e.g., IVIVC or PBAM), or the dissolutionmethod and resulting dissolution profiles (i.e., fastestand slowest profiles) may provide a safe space withinwhich similar clinical performance has beendemonstrated.

Breakout session subgroups B1 and B2 deliberatedseveral specific questions related to bridging Bbiorelevant^and QC methods:

1. Is it ever feasible to use a biorelevant method for QC?Is there any rationale to implement two differentmethods (QC and biorelevant)? How does one linkbiorelevant testing to QC testing during productdevelopment?a. Dissolution performed in SGF using USP Appa-

ratus 1 or 2 for BCS 1 and 3 drug products can beconsidered biorelevant. Thus, for these drug products,the QC method can be considered Bbiorelevant.^Using biorelevant dissolution methods for poorlysoluble drug products in a QC environment ischallenging mainly due to concerns over lack ofmethod robustness, not achieving full drug product

Fig. 6. Fishbone diagram of input factors that may impact in vitro dissolution


release (e.g., Q of 80%), and general global regulatoryacceptance.b. Biorelevant methods are mainly used to rank

order formulation prototypes and to identify potentialbiopharmaceutics risk. They can also be used toidentify CMAs and CPPs early in development. Inpractice, companies perform biorelevant and QC typedissolution in development and then switch to a QCmethod in late stage development and use informa-tion from biorelevant dissolution to establish appro-priate discrimination of the QC method toward CPPsand CMAs.

2. Is it critical that the extent of dissolution meet a Q of80%? There’s a tension between biorelevance andcomplete release for poorly soluble products.a. Incomplete dissolution profiles are likely

concerning from a regulatory perspective and maybe called into question if these profiles are inconsis-tent with the systemic concentration–time profile ofthe product. Incomplete dissolution maybe acceptableprovided that the sponsor provides additionalsupporting data (i.e., from BE studies or based onIVIVC).b. Adding surfactant to the dissolution medium to

achieve full release on the other hand may sacrificediscriminating power of the dissolution method to-ward CPPs and CMAs.

3. In your experience, do you mostly use biorelevantdissolution as tool in early stage drug development? Doyou have comments on additional roles of biorelevantdissolution?a. Yes, biorelevant dissolution is mainly applied in

early development.b. Some companies may use biorelevant dissolution

in decision-making related to formulation and process

changes and as input toward bioequivalence studydesign.

4. How do you determine the appropriate level ofdiscrimination for biorelevant, clinically relevant, andQC dissolution methods?a. Clinical relevance of the dissolution method

implies that a correlation between CMAs and CPPsexists. The magnitude of in vitro discriminationshould ideally be aligned with in vivo differences.Therefore, the method is considered appropriatelydiscriminating with regard to rejecting batches thatare not bioequivalent to batches used in pivotalclinical trials. In the case of an established safe space,the dissolution method conditions may be overlydiscriminating.b. Biorelevant dissolution methods used in early

product development often follow a Bgeneric^ ap-proach. Therefore, in the absence of a link to in vivoperformance, the degree of discriminating power isoften not known. Biopharmaceutics risk assessmentand prior knowledge, as well as modeling andsimulation, may be helpful to inform the need andguide approaches to adjust the sensitivities towardcertain CMAs (i.e., API particle size) or CPPs.c. The discriminating power of a QC method may be

established based on a biopharmaceutics risk assess-ment aimed at identifying CPPs and CMAs. In theabsence of clinical relevance, the appropriate magni-tude and significance of discrimination is unclear.

5. If direct correlations such as those shown in the Day 1afternoon presentations (22,24,26) show a direct link ofprocess parameters and materials attributes and PK,what is the value of in vitro dissolution? What if theseparameters or material attributes are a better predictorof in vivo performance?

Table II. PAT Technologies, Applied RTRT Methods, and Targeted Materials for Measurements

PAT technology RTRT method Material measured

Laser diffraction Granule particle size Milled granulesNIR API content, water content Final blendWeight, thickness, hardness Tablet weight, hardness, thickness Core tablet

Fig. 7. Predicted (solid line) and measured dissolution rates for batches 1 and 2 demonstrating theagreement of measured and modeled data


Development of dissolution models and surrogatetesting should be performed in parallel with aclinically relevant QC method. Once a clear correla-tion between the surrogate test(s), dissolution, andin vivo performance is established, surrogate testingcan be used.

Detailed outcomes of the breakout sessions are providedin the Supplemental Material.

CONCLUSIONS

In vitro dissolution testing continues to play a major rolein drug product development, quality control, and to supportpostproduct approval formulation and manufacturingchanges. Today, scientists have increasing access to in vitrotools which can be used to build a link to in vivo performanceand to guide formulation candidate selection, formulationprocess development and to assess and respond tobiopharmaceutics risks including informing the success ofnecessary formulation bridging studies and ultimately for theroutine release of product. The new and expanding set ofdiagnostic tools which includes modeling and simulation forboth in vitro dissolution and PBAM also creates a dilemmafor the industry and regulators. Both parties want to ensurethat product released to patients meets the claims in theproduct label; however, the lack of robustness of nontradi-tional methods used for example in early development oftenpresents a challenge in utilizing those in a routine QCenvironment. Product development is usually performed byhighly skilled scientists who are competent to work with noveltechnology and trained to interpret development data in thecontext of biopharmaceutics risks. Transferring nontraditionalmethods to a routine QC environment with strict adherenceto procedures and where data are usually judged Bpass^ orBfail^ is concerning, as failing results due to methodvariability are likely to cause major disruptions in the releaseof product and thus supply of product to patients.

One of the key objectives of the Day 1 discussions was toengage scientists in productive dialog to share experiences,new opportunities, and concerns over the current state ofproduct understanding especially in the context of usingdissolution testing as an integral component of robust controlstrategies to ensure product quality and to enable emergingformulation and manufacturing technology. In this respect,the technical presentations provided excellent backgroundinformation, while the breakout sessions highlighted majordifferences in the use of dissolution-related terminology,utilization and the limitations of novel approaches, andinterpretation of regulatory guidance. As the discussionsduring the breakout sessions indicated, the meeting was asuccess with respect to gaining a better understanding ofterminology, the potential use of dissolution technologyincluding surrogates and modeling to enhance productunderstanding, and the need to develop clinically relevantdissolution specifications especially for poorly soluble drugs.The challenges to overcome current limitations with theavailable in vitro toolkit and with novel approaches such asPBAM are part of the broader effort to develop strategies toimplement CRDPS in product development and, if appropri-ate, as part of the control strategy culminating in RTRT and

the desired regulatory flexibility for product life-cycle man-agement. Details about the current state and challengesrelated to PBAM and establishing traditional IVIVC as wellas strategies to develop CRDPS are detailed in Day 2 andDay 3 meeting summary reports.

Clearly, one of the shortcomings of Day 1 was the limitedtime that was set aside for the breakout sessions with respectto the significance and the number of topics that werediscussed. It may be advantageous for future workshops todedicate considerably more face-to-face time to discuss topicsthat concern scientists working in the industry and regulatoryagencies to advance product understanding as it relates todissolution testing for patient benefit.

ACKNOWLEDGMENTS

The meeting organizers are indefinitely grateful to Drs.James Polli (University of Maryland, School of Pharmacy,Baltimore, MD) and Ms. Ann Anonson (UM) for theirtremendous efforts in helping in the organization of thisworkshop.

COMPLIANCE WITH ETHICAL STANDARDS

Conflict of Interest The authors declare that they have noconflicts of interest.

Publisher’s Note Springer Nature remains neutral with regard tojurisdictional claims in published maps and institutional affiliations.

REFERENCES

1. Abend A, Heimbach T, Cohen M, Kesisoglou F, Pepin X,Suarez-Sharp S. Dissolution and translational modeling strate-gies enabling patient-centric drug product development: the M-CERSI workshop summary report. 2018. p 60.

2. Heimbach T, Kesisoglou F, Suarez S, Kakhi M, Nico HolmstockN, Olivares-Morales A, et al. Dissolution and translationalmodeling strategies enabling patient-centric drug product de-velopment, M- CERSI workshop summary – establishing invitro in vivo link using PBPK absorption modeling. AAPS J.submitted.

3. Suarez-Sharp S, Cohen M, Kesisoglou F, Abend A, Marroum P,Delvadia P, et al. Applications of clinically relevant dissolutiontesting: workshop summary report. AAPS J. 2018;20(6):93.

4. Dokoumetzidis A, Macheras P. A century of dissolutionresearch: from Noyes and Whitney to the BiopharmaceuticsClassification System. Int J Pharm. 2006;321(1):1–11.

5. United States Pharmacopeia and the National Formulary (USP-NF). Chapter <711> Dissolution. Rockville: US PharmacopeialConvention; 2015.

6. Yu LX. Pharmaceutical quality by design: product and processdevelopment, understanding, and control. Pharm Res.2008;25:781–91.

7. Yu LX, Amidon G, Khan MA, Hoag SW, Polli J, Raju GK,et al. Understanding pharmaceutical quality by design. AAPS J.2014;16(4):771–83.

8. Koenigsknecht MJ, Baker JR, Wen B, Frances A, Zhang H, YuA, et al. In vivo dissolution and systemic absorption ofimmediate release ibuprofen in human gastrointestinal tractunder fed and fasted conditions. Mol Pharm. 2017;14(12):4295–304.


9. U.S. Department of Health and Human Services - Food andDrug Administration; Center for Drug Evaluation and Re-search (CDER). Waiver of in vivo bioavailability and bioequiv-alence studies for immediate-release solid oral dosage formsbased on a Biopharmaceutics Classification System—guidancefor industry. Silver Spring: FDA; 2017.

10. Miksinski SP. Clinical relevance—seeing the big picture. InDissolution and translational modeling strategies enablingpatient-centric product development, Baltimore, MD. 2017.

11. Miksinski SP. New drug applications: the evolution of qualityreview. In 2nd FDA/PQRI conference on advancing productquality, Bethesda, MD. 2015.

12. Ju R, Grady H. Industry perspective on the current status andfuture of dissolution testing for product development andquality control. In Dissolution and translational modelingstrategies enabling patient-centric product development, Balti-more, MD. 2017.

13. Mann J, Dressman J, Rosenblatt K, Ashworth L, Muenster U,Frank K, et al. Validation of dissolution testing with biorelevantmedia: an OrBiTo study. Mol Pharm. 2017;14(12):4192–201.

14. Kuiper J, Coutant C. Use of bio-predictive methods during earlyformulation screening with case studies. In Dissolution andtranslational modeling strategies enabling patient-centric prod-uct development, Baltimore, MD. 2017.

15. Sugano K, Okazaki A, Sugimoto S, Tavornvipas S, Omura A,Mano T. Solubility and dissolution profile assessment in drugdelivery. Drug Metab Pharmacokinet. 2007;22(4):225–54.

16. Kuiper J, Harmon P. Dissolution rate limiting AUC: simplemethodology for measuring dissolution rate of the entire dose inbiorelevant media. In AAPS annual meeting, San Diego, CA,USA. 2014.

17. Psachoulias D, Vertzoni M, Goumas K, Kalioras V, Beato S,Butler J, et al. Precipitation in and supersaturation of contentsof the upper small intestine after administration of two weakbases to fasted adults. Pharm Res. 2011;28:3145–58.

18. Bhattachar SN, Perkins EJ, Tan JS, Burns LJ. Effect of gastricpH on the pharmacokinetics of a bcs class II compound in dogs:utilization of an artificial stomach and duodenum dissolution

model and gastroplus simulations to predict absorption. J PharmSci. 2011;100(11):4756–65.

19. Lu X, Han J-H, Mattocks D. Dissolution methodologies frombiorelevant to quality control: challenges and gaps. In Dissolu-tion and translational modeling strategies enabling patient-centric product development, Baltimore, MD. 2017.

20. Wang Q, Fotaki N, Mao Y. Biorelevant dissolution: methodol-ogy and application in drug development. Diss Technol.2009;16(3).

21. Blanquet SP, Zeijdner E, Beyssac E, Meunier J-P, Denis S,Havenaar R, et al. A dynamic artificial gastrointestinal systemfor studying the behavior of orally administered drug dosageforms under various physiological conditions. Pharm Res.2004;21:585–91.

22. Zhang L, Hermans A. The use of surrogates for dissolutiontesting for IR formulations: when is it feasible?. In Dissolutionand translational modeling strategies enabling patient-centricproduct development, Baltimore, MD. 2017.

23. International Conference on Harmonization of Technical Re-quirements for Registration of Pharmaceuticals for Human use(ICH). ICH Harmonized Tripartite Guideline: Specifications:Test Procedures and Acceptance Criteria for new DrugSubstances and new Drug Products: Chemical Substances,Q6A,. In 1999.

24. Drazer G, Pawar P, Wang Y, Keyvan G, Callegari G, Cuitino A,Muzzio F. Enabling real time release testing (RTRt) with NIR-based prediction of dissolution for tablets made by continuousdirect compression (CDC). In Dissolution and translationalmodeling strategies enabling patient-centric product develop-ment, Baltimore, MD. 2017.

25. Pawar P, Wang Y, Keyvan G, Callegari G, Cuitino A, Muzzio F.Enabling real time release testing by NIR prediction ofdissolution of tablets made by continuous direct compression(CDC). Int J Pharm. 2016;512(1):96–107.

26. Li H, Prichard J, Swinney KA. Dissolution modeling for realtime release testing (RTRT). In Dissolution and translationalmodeling strategies enabling patient-centric product develop-ment, Baltimore, MD 2017.


Documents

Meeting Report Dissolution Testing in Drug Product ... · product release and stability testing are highlighted in the ﬁrst two summaries of the Day 1 podium presentations. Additionally,