`DOI: 10.1007/s11095-008-9822-x
`
`Expert Review
`
`The Science of USP 1 and 2 Dissolution: Present Challenges
`and Future Relevance
`
`Vivian Gray,1,7 Gregg Kelly,2 Min Xia,3 Chris Butler,4 Saji Thomas,5 and Stephen Mayock6
`
`Received September 4, 2008; accepted December 24, 2008; published online January 23, 2009
`Abstract. Since its inception, the dissolution test has come under increasing levels of scrutiny regarding its
`relevance, especially to the correlation of results to levels of drug in blood. The technique is discussed,
`limited to solid oral dosage forms, beginning with the scientific origins of the dissolution test, followed by
`a discussion of the roles of dissolution in product development, consistent batch manufacture (QC
`release), and stability testing. The ultimate role of dissolution testing, “to have the results correlated to in
`vivo results or in vivo in vitro correlation,” is reviewed. The recent debate on mechanical calibration
`versus performance testing using USP calibrator tablets is presented, followed by a discussion of
`variability and hydrodynamics of USP Apparatus 1 and Apparatus 2. Finally, the future of dissolution
`testing is discussed in terms of new initiatives in the industry such as quality by design (QbD), process
`analytical technology (PAT), and design of experiments (DOE).
`in vitro–in vivo correlation; quality by design;
`
`KEY WORDS: biorelevant methods; dissolution;
`variability.
`
`INTRODUCTION
`
`This paper explores the advantages and disadvantages of
`the current methodology in light of recent challenges. While
`acknowledging its limitations, a case is made that the current
`dissolution test for drug product performance has value.
`The scope of this paper includes information on current
`issues, but it is not a tutorial on dissolution testing. The focus
`is on USP Apparatus 1 (baskets) and 2 (paddles) because
`these two systems constitute the bulk of dissolution testing in
`the pharmaceutical industry (1).
`The paper is organized by contemporary dissolution topics.
`Presented first is a description of the current challenges the paper
`will address. The challenges generally are divided into two
`classes, biorelevance and variability. Challenges covered by each
`subheading are discussed, followed by a brief section on the
`origin of the method procedure governed by United States
`Pharmacopeia chapter on Dissolution <711> (1). The intent is
`to demonstrate the scientific basis of current industry practice.
`Then, a review of dissolution by application exposes both the
`
`1 V. A. Gray Consulting, Inc., 9 Yorkridge Trail, Hockessin, Delaware
`19707, USA.
`2 Analytical R & D, Pfizer Global R & D, Groton, Connecticut, USA.
`3 Analytical Development, Vertex Pharmaceuticals, Cambridge, Mas-
`sachusetts, USA.
`4 Validation and CAPA, Ortho Clinical Diagnostics, Rochester, New
`York, USA.
`5 QC Lab Operations, Par Pharmaceutical, Spring Valley, New York,
`USA.
`6 Analytical Services, Catalent Pharma Solutions, Research Triangle
`Park, San Diego, North Carolina, USA.
`7 To whom correspondence should be addressed. (e-mail: vagray@rcn.
`com)
`
`value and limitations of the technique as an analytical tool.
`Application to formulation development, quality control, and in
`vitro–in vivo correlations (IVIVC) is covered. Next, variability
`inherent to dissolution testing is explored in the context of the
`challenges. A discussion is presented on calibration, including use
`of physical measurements and calibrator tablets, plus error
`associated with experimental conditions or analyst technique. It
`should be noted that recently the calibrator tablets were renamed
`by USP as Performance Verification Standards; however, since
`this is a new development, we have kept the term calibrator
`tablets throughout this paper. Hydrodynamics is the final section
`under dissolution method variability. The future of dissolution
`testing is discussed in sections on process analytical testing (PAT),
`design of experiments (DOE), and quality by design (QbD).
`Finally, the utility and future of the technique are summarized.
`
`Challenges
`
`Both biorelevance and technique variability are used to
`challenge the validity of dissolution testing. The basis for each
`challenge is presented below.
`The most significant challenge for many dissolution
`methods used as a nominal performance measure stems from
`the lack of biorelevance. Scientists have stated that develop-
`ing a dissolution method and setting associated specifications
`that are not linked to in vivo performance may limit the value
`of testing (2–7). It is not difficult to see that the vortex in the
`current design of USP apparatus is not the same as in a
`churning stomach. The majority of dissolution testing is
`carried out in a simple salt medium at a particular pH. The
`gastrointestinal lumen is significantly different, containing a
`plethora of biomolecules and salts in a changing pH
`
`1289
`
`0724-8741/09/0600-1289/0 # 2009 Springer Science + Business Media, LLC
`
`ENDO - Ex. 2028
`Amneal v. Endo
`IPR2014-00360
`
`
`
`1290
`
`Gray et al.
`
`environment. A lack of a biorelevant (physiologically based)
`dissolution system and specification often leads to data that
`are disconnected from in vivo results (2,3,7,8). Examples are
`cited where the dissolution method is either overly or not
`sufficiently discriminating (2,7,9). Few cases have been found
`where the method is appropriately discriminating (10). A
`dissolution method that is developed solely as a quality
`control tool for manufacturing is much less desirable than one
`that has bearing on patient safety or efficacy. If measurements
`have no bearing on the pharmacokinetic impact, then testing
`is not controlling the most important aspect of performance
`(2,3,7,8). Calls for better method development with biorele-
`vant specifications are the result (2,3,6,7).
`Variability associated with dissolution testing is another
`area receiving a great deal of attention. Many studies
`demonstrate the source and extent of variability (2–4,8,11–
`14). These sources can be divided into four subsets. The first is
`the physical or mechanical setup of the test. Tolerances allowed
`in operating the apparatus are defined by the USP (1). The
`definitions are designed to allow the apparatus to function with
`acceptable method variability, but even when operating within
`these limits, different dissolution profiles for the same drug
`product may result. Other physical factors are not controlled
`by the USP description but have an effect. Among the
`parameters in this class are shaft or basket wobble, vessel/shaft
`tilt, shaft centering, shaft height in vessel, and rotational speed
`(3,13). Vessel roundness, surface uniformity, or other hydrody-
`into this class and impact results (3–
`namic effects fall
`5,11,14,15). Even small changes in basket mesh size seem to
`have an influence on results (14). Another class of variability
`arises from operational differences. Parameters in this group
`are incidental vibration, the extent of degassing, inconsistent
`tablet placement in vessels, and inconsistent use of clips or
`sinkers (3,11–13). The third class of variability comes indirectly
`from performance differences in calibrator tablets that are real
`(8,16,17), operator induced, or from excipient deposition (18).
`As the name implies, calibrator tablets are used to verify
`overall system precision to qualify apparatus and control
`system variability. However, different disintegration mecha-
`nisms between calibrator and sample tablets are cited as a
`source of variability (3). Proposed remedies for calibrator
`tablet variability are mechanical calibration (3,13,19), project-
`specific manufacturer calibrator tablets possessing similar
`processing and mechanistic disintegration qualities (3,6), or
`non-USP apparatus (4,5,20). The fourth source of variability
`comes from manufacturing and is due to lot-to-lot or tablet-to-
`tablet processing or handling differences of the drug product
`(3,6,16). It includes particle size distribution and polymorph
`changes during drug substance manufacture. Changes in
`excipient characteristics are known to impact results (7). The
`variability from this cause is independent of the method but is
`reflected in the results. Sorting out the origin among all the
`potential sources of variability can be problematic.
`
`Scientific Origins of Dissolution
`
`Scientific Origins
`
`determining that the dosage would dissolve. To this day,
`dissolution is the only test that indicates if a dosage form will
`dissolve in the patient. The disintegration test was the first
`test designed to do this, but it has obvious limitations.
`Although a tablet or capsule can disintegrate into smaller
`particles,
`if it does not dissolve,
`it is not available to be
`absorbed in the small intestine.
`Dissolution, as a general dosage performance test, was
`primarily linked to changes in the drug product formulation
`and the critical process parameters that can affect dissolution.
`During the process validation of
`tablet manufacturing,
`dissolution testing is performed on tablets at
`the target
`hardness and at the high and low extremes.
`Dissolution is still a critical test to determine the effects of
`aging of the product on stability. Changes in tablet hardness,
`moisture, or other excipient changes can affect dissolution.
`Capsule cross-linking can have a significant effect on dissolution
`of samples on stability. In many respects, this continues to be the
`most compelling reason to have an effective dissolution test for
`testing a solid oral dosage product.
`Some of the basic aspects of the dissolution test have
`their origins in general conditions in the human body. The
`test is conducted at 37°C. The paddle or basket rotation is
`designed to produce reproducible hydrodynamics that can be
`consistent from lab to lab. The real physical purpose of the
`agitation is to remove the drug-saturated layer of dissolution
`from around the dosage and replace it with fresh medium
`without causing a significant physical change in the dosage.
`The use of a 900-mL volume was determined in order to be
`enough to establish sink conditions (at least three times
`saturation) for most active pharmaceutical ingredients. Dis-
`solution media were developed to mimic the pH of the gastro-
`intestinal tract. At one time, simulated intestinal fluid had a
`pH of 7.4. This was changed to a pH of 6.8 in the mid-90s,
`because it was determined that this more closely represents
`the intestinal pH (21).
`
`Dissolution Testing within the USP
`
`The basic dissolution test in USP chapter <711> Disso-
`lution describes the apparatus, the dissolution procedure, and
`product specifications. The old chapter <724> described
`Apparatus 3 through 7, while Apparatus 1 and 2 were
`described in chapter <711>. The newer editions of the USP
`have now combined Apparatus 1 through 4 in chapter <711>.
`The dissolution procedures have been harmonized in the
`pharmacopeias internationally, although there are some
`sections that remain unique to each pharmacopeia. The
`USP chapter <1088> describes the procedure for in vitro–in
`vivo evaluation of dosage forms, and chapter <1092> presents
`the development and validation of the dissolution procedure.
`The dissolution test has evolved over time and will continue
`to be improved as it is called upon to give more data that are
`relevant to dosage performance in the patient.
`
`The FDA and Dissolution Testing in History
`
`Initially, the dissolution test was used primarily as a
`formulation development tool and as a quality control test for
`
`The FDA has placed much importance on the dissolution
`test and reviews the USP monograph dissolution tests for
`consistency with the dissolution conditions in the approved
`
`
`
`USP 1 and 2 Dissolution: Present Challenges and Future Relevance
`
`1291
`
`product’s New Drug Application. Most solid oral dosage
`forms are required to have a dissolution test, and it is not
`uncommon to have a drug recall due to a failed dissolution
`test. Part of the approval process of an NDA solid oral
`dosage is the FDA evaluation of the dissolution method.
`Members of the FDA helped to develop the Biopharma-
`ceutical Classification System (BCS). This has led to the
`development of guidances related to dissolution testing that
`are available on the FDA website.
`
`Utility and Basic Goals of the Test—Formulation Development
`
`Formulators consider the dissolution test to be a very
`powerful tool. The test can be used to show the dependence
`of dissolution rate on the presence and concentration of
`certain excipients and on manufacturing variables. There is
`abundant literature on the use of dissolution as a comparative
`test that can show a formulation change. A select few will be
`highlighted in this section of the paper.
`As early as 1976, Khan and Rooke (22) described the effect
`of disintegration type upon the relationship between compres-
`sion force and dissolution efficiency. The discussion compared
`three common disintegrants, sodium carboxymethyl cellulose,
`sodium starch glycolate, and a cation-exchange resin to then less
`commonly known excipients as insoluble sodium carboxymethyl
`cellulose, casein formaldehyde, calcium carboxymethyl cellu-
`lose, and a cross-linked polyvinylpyrrolidone. They concluded
`that the disintegrant type has a pronounced effect on the
`dissolution rate. In this work, the paddle was used at 50 rpm with
`water as the medium. The study also cited a 1963 publication of
`Levy et al. (23) that correlated an increase in compression force
`with starch-containing formulations.
`Chowhan and Palagyi (24) explored the issues of
`hardness and the effect of the dissolution rate. This study
`showed how hardness was increased by partial moisture loss
`in compressed tablets. Several
`factors were investigated
`including type and percentage of excipient, water solubility,
`hygroscopicity of excipients or drug, and the influence of
`frequently used binders. They concluded that since the
`dissolution is related to moisture content of the granulation
`and the hardness of the tablets at the time of compression, the
`dissolution specification would ensure that the tablets meet the
`moisture and hardness requirements. It was recommended
`that the moisture content of the granulation and initial
`hardness be used as in-process controls. In this work, the
`paddle was used at 120 rpm with water or 7.4 phosphate buffer
`as the medium.
`In 1981 Taborsky-Urdinola et al. (25) published a paper
`that won an APHA Research award. The importance of the
`paper was the proof that packaging type and storage conditions
`in multiple and unit dose containers markedly affect the
`dissolution results of model Prednisone tablets. A conclusion
`was that relabeling repackaged tablets with the expiration date
`of the original container was invalid. The dissolution conditions
`were paddle at 50 rpm using water as the medium.
`Chowhan and Chi (26) continued his research and in
`1985 described the role of lubricants and their effect on
`dissolution results. Two lubricants, magnesium stearate and
`sodium stearyl fumarate, were compared under identical mixing
`conditions to determine drug–excipient interactions. The con-
`clusions were that sodium stearyl fumarate did not exhibit drug–
`
`excipient interactions, whereas magnesium stearate did exhibit
`significant drug–excipient interactions that adversely affected
`the disintegration time and dissolution rate.
`Changes in surface area and dissolution rate were illumi-
`nated by Sunada et al. (27) in 1989. The changes in surface area
`during the dissolution process were measured, and the relation-
`ship between the surface producing rate constant and the initial
`particle size of sieved samples was estimated. There was also the
`simulation of the dissolution process based on the changes in
`surface area and the surface producing rate constant. The
`paddle speed was 250 rpm in water.
`In the 1990s, the use of dissolution as an indicator of
`aging began. Chowhan (28) discussed the complexity of aging
`as related to selected factors other than the packaging and
`storage conditions. Factors such as the hygroscopicity of the
`superdisintegrants; method of disintegrant incorporation; gran-
`ulation moisture content; effect of high or low humidity on the
`type of disintegrant (e.g., dibasic calcium phosphate dihydrate
`and tribasic calcium phosphate); effect of the use of lactose,
`dextrose, or MMC; and gelatin shell cross-linking were all
`evaluated. It was concluded that guidelines calling for acceler-
`ated conditions could give a good indication of aging issues.
`Babu and Pandit (29) described how stability of gliben-
`clamide was enhanced by complexation with β-cyclodextrin.
`The dissolution rate was employed as an indicator of aging
`using the paddle at 100 rpm and pH 7.4 phosphate buffer.
`Dissolution rate was one of the important parameters
`measured when differentiating forms I and II (R, S) of
`propranolol hydrochloride. Bartolomei et al. (30) showed that
`dissolution rates of the two polymorphs were different using
`the paddle at 50 rpm with 0.1 N hydrochloric acid medium;
`the test was run at 20°C and 37°C.
`The effects of temperature and humidity on the physical
`properties of piroxicam tablets were shown by Sarisuta et al.
`(31). The tablets containing various fillers (lactose or
`mannitol) were studied after storage for 12 weeks at 40°C
`and 52% relative humidity (RH) and 40°C and 96% RH. The
`physical properties of
`the tablets were measured every
`2 weeks. Dissolution was measured using the paddle at
`50 rpm with simulated gastric fluid as the medium. The
`dissolution rate decreased from week to week, regardless of
`the filler used. It was explained that the decrease in dissolution
`was due to moisture sorption by the tablet ingredients, which
`led to the formation of a saturated solution of water-soluble
`substances. Consequently, crystal growth and swelling of
`polymeric material occurred. This yielded a continuous
`structure of larger crystals so the exposed surface area was
`significantly reduced, hence the dissolution rate decreased.
`In 1999 Rohrs et al. (32) showed the effect of croscar-
`mellose sodium disintegrant on delavirdine mesylate. In the
`presence of high humidity, the water presumably acted as a
`reaction medium and a plasticizer for croscarmellose sodium,
`facilitating protonation of the carbonyl sites on the disintegrant.
`The important finding is that this reaction could very well occur
`with any acid salt of a free base. A change in inter-particle
`bonding can explain the reduction in tablet deaggregation
`during dissolution. The dissolution was performed using the
`paddle at 50 rpm, and the medium was 0.05 M phosphate buffer
`at pH 6 with 0.6% sodium dodecylsulfate surfactant.
`The effect of powder substrate composition on the
`dissolution rate of methyclothiazide liquisolid compacts was
`
`
`
`1292
`
`Gray et al.
`
`illustrated by Spireas et al. (33). Dissolution rates were
`increased by optimizing carrier-to-coating ratios in methyclo-
`thiazide liquisolid tablets containing a 5% w/w drug solution
`in polyethylene glycol 400 with difference excipient ratios.
`The dissolution conditions used were the paddle at 50 rpm
`and a medium of 0.1 N hydrochloric acid.
`The influence of excipients, especially binders, on the
`dissolution rate of paracetamol tablet formulations was shown
`by Abebayo et al. (34). The effect of binders, namely
`breadfruit and cocoyam starch mucilage binders, was related
`to their surface tension and viscosity. The dissolution test
`used the basket at 50 rpm and pH 5.8 phosphate buffer.
`Numerous articles on the subject of extended-release
`formulation information show that there is variability in the
`dissolution rate with the change of matrix ingredients and
`ratios (35–38). In the references cited, as in many cases, the
`paddle was used in the dissolution test.
`In the last 30–40 years, the dissolution test with USP
`paddle and basket apparatus has been used extensively to
`provide information to the formulators regarding critical
`process variables. Only a limited amount of the literature is
`shown here as the literature is full of examples of in vitro
`release testing used to determine change in the formulation
`or manufacturing process. The power of the dissolution test is
`undisputed in assisting product development
`from early
`phases to monitoring stability.
`
`QC Testing for Batch Manufacturing Consistency
`and Specification Setting
`
`Product Batch Release
`
`The value of in vitro dissolution testing as a quality
`control tool is demonstrated by its long history of regulatory
`acceptance. Dissolution testing has been included in the USP
`since 1970 and continues to be an important test today as
`evidenced by the large number of monographs that include
`dissolution requirements (over 600 as of 2006) (39). This
`points to an important benefit for drug manufacturers—
`dissolution testing fulfills a regulatory requirement.
`Although the primary purpose of the dissolution test
`specification is to distinguish between acceptable and unaccept-
`able batches, it is also used as a measure of batch-to-batch
`consistency of the manufacturing process. In this case, the
`method may be developed to be sensitive to manufacturing
`variables determined to influence drug release (40).
`
`Stability and Shelf Life
`
`Dissolution testing is also the primary method used to
`demonstrate stability of drug product performance through-
`out its shelf life. Although not specified by name in the
`guidance, dissolution testing fulfills the ICH Q1A (R2)
`requirement that stability studies include testing of drug
`influence “product quality, safety and/or
`attributes that
`efficacy” and that are susceptible to change over time (41).
`Dissolution has proven to be a valuable tool to indicate
`changes in such characteristics as crystallinity (42), glass
`transition temperature and pore structure of polymeric
`excipients (43), polymorphism (44), gelatin capsule cross-
`
`linking (45), and moisture content (32). This information can
`be used to make informed decisions on selection of formula-
`tion, manufacturing process, and packaging.
`
`Setting Specifications, Establishing Product History,
`Post-Approval Manufacturing Changes
`
`The dissolution test plays an important role in setting
`drug product specifications. The dissolution specification
`includes the specific dissolution procedure as well as accep-
`tance criteria;
`it is intended to show that manufactured
`product is bioequivalent to pivotal clinical lots and confirm
`it was manufactured within acceptable values of critical
`manufacturing variables. Conformance to the acceptance
`criteria can be used to determine stability of the drug and to
`justify waiver of additional clinical studies following certain
`post-approval changes (46–48). Following approval, dissolu-
`tion data for manufactured lots form a product history from
`which the “true” capability and variability of the process can
`be derived. This information may be used as justification for
`revised acceptance criteria (49).
`Recently, the dissolution test has been criticized for not
`being predictive of bioavailability because methods do not
`mimic GI conditions closely enough (45). The lack of
`predictivity is not necessarily a limitation of the test, but
`may result from inappropriate selection of acceptance criteria
`or specific analytical conditions. In some applications, an
`overly sensitive dissolution test is desirable. For example, the
`FDA guidance on dissolution testing of immediate-release
`(IR) solid oral drugs (40) includes a procedure for manufac-
`turing bioequivalent product
`lots with different in vitro
`dissolution to identify and establish an acceptable range for
`critical manufacturing variables. As for the question of
`biorelevance, because the goal of the dissolution procedure
`is to establish equivalence with acceptable clinical lots, the
`procedure need only be predictive of bioavailability. For this
`purpose, mimicking the gastrointestinal tract is not relevant.
`Ensuring that the procedure is predictive should be addressed
`during a rational method development
`following QbD
`principles.
`
`Tests for Similarity and Difference
`
`Because the comparison of dissolution profiles is used to
`evaluate the effects of formulation changes, the stability of
`product performance over time, and lot-to-lot manufacturing
`consistency and to demonstrate bioequivalence, it is important
`to understand the strengths and weaknesses of the various
`methods used for comparing them. Coming up with an
`objective data-based means of deciding if dissolution profiles
`are similar or different is a challenge. Because a dissolution
`profile is a plot of cumulative percent drug released (i.e., each
`data point is dependent on the previous data point) versus
`time, the underlying assumption of data independence is
`violated, precluding the use of statistical tests of difference
`(50). The use of exploratory data analysis methods, such as
`overlapping confidence intervals at individual time points as a
`test of similarity, becomes problematic when they overlap at
`some, but not all, of the time points.
`
`
`
`USP 1 and 2 Dissolution: Present Challenges and Future Relevance
`
`1293
`
`Mathematical comparison methods such as f1 and f2
`(51,52) utilize differences between average reference and test
`profiles at each sampling interval to provide a single number
`with which to quantify the similarity or difference between
`them. The mathematical comparator f2 (similarity factor)
`(51), which is recommended by the FDA (40,46–48), has the
`advantage of being easy to calculate. However, this technique
`is sensitive to the number of dissolution time points after the
`plateau is reached (52) and does not account for vessel-to-
`vessel variability, and there does not appear to be a well-
`defined basis for the “sameness” threshold of f2=50 (50). More
`importantly, because f2 is a sample statistic and is not based on
`a known population, the probability of type I (rejecting similar
`profiles as dissimilar) and type II (accepting dissimilar profiles
`as being similar) error is unknown (50,53,54). A modification
`to this method, in which f2 is calculated for each individual
`dosage unit, has been used to allow the inter-vessel variability
`to be expressed (54). The use of bootstrapping to simulate
`confidence intervals has been used with highly variable data to
`avoid making false conclusions (52,55).
`Model-dependent methods involve fitting the reference
`dissolution profile data to a mathematical function (known
`physical curve); similarity of a test profile is evaluated in
`terms of difference between the mean model parameters of
`the reference and test curves (56). Models that have been
`used include zero-order (54,56), first-order (54,57), Hixson–
`Crowell (54,57), Higuchi (54,57), quadratic (57), Weibull
`(54,56–58), Gompertz (57,58), Probit (58), exponential (58),
`and logistic (57,58). These methods have the advantage of
`taking into account variance and covariance of the data sets,
`and sampling time points for the reference and test profiles
`do not have to be the same. However, it is not always possible
`to find a model that adequately fits the data. Selection of an
`inappropriate model curve can yield misleading results,
`resulting in incorrect conclusions, so it is important to run a
`lack-of-fit test on the reference data prior to comparing
`model parameters (59).
`Statistical multivariate methods using multivariate
`ANOVA have also been used (60,61). These do take into
`account variability and correlation structure of cumulative
`percent-released-versus-time data. An advantage is that they
`can be used to make estimates of type I and type II errors.
`
`In Vitro and In Vivo Relationships and Bioequivalence
`Challenges in Dissolution Method Development
`
`IVIVCs were introduced as the desire of both industry
`and regulatory agency to reduce development time, cost, and
`regulatory burden (62). Recognizing that dissolution rate,
`aqueous solubility, and gastrointestinal permeability are the
`key parameters that control the rate and extent of drug
`absorption, Amidon et al. (63) proposed a Biopharmaceutics
`Classification Scheme (BCS) in 1995. Later, FDA classified
`drug substances into four groups: class I—high solubility, high
`permeability; class II—low solubility, high permeability; class
`III—high solubility, low permeability; class IV—low solubil-
`ity, low permeability (48,64). For rapidly dissolving class I
`drugs, because of their high solubility and high permeability
`characteristics, the in vivo dissolution is not the rate-limiting
`step, so IVIVC may not be possible (63–65). In addition,
`since gastric emptying is the key factor in determining the
`
`plasma profile, if the excipients in the drug product alter the
`gastric-emptying rate, bioinequivalent products will be the
`result. For class II drugs, on other hand, dissolution may be
`the limiting step of the drug absorption, therefore, an IVIVC
`may be expected (21,62). More research is needed to develop
`and validate in vitro dissolution methods for class II drugs so
`that they can be used to predict in vivo dissolution (66). For
`class III drugs, permeability is the limiting step of
`the
`absorption, and a limited IVIVC may be expected, and
`finally, for class IV drugs, IVIVC is difficult. The drug will
`have both limited dissolution and permeability so it will be
`difficult, at best, to develop a dissolution model unless the
`permeability is borderline low.
`Currently, there are four levels of IVIVC defined in
`FDA guidances (62,67–72). Level A correlation is a point-to-
`point relationship between in vitro dissolution and the in vivo
`pharmacokinetic data (73). It
`is generally linear and is
`reviewed as a predictive and preferred approach (62,73,74).
`In the case of a level A correlation, in vitro dissolution data
`can serve as a surrogate for in vivo performance. For a class I
`drug, IVIVC is generally not likely, but when formulated as
`an extended-release product and the solubility and perme-
`ability of the drug is site-independent, a level A correlation is
`expected (75,76). For a class II drug formulated as an
`extended-release product, and the solubility and permeability
`of the drug are site-independent, a level A correlation is also
`likely; however, if the permeability is site-dependent, IVIVC
`is unlikely (75). Level B correlation applies the principles of
`statistical moment analysis. It compares the mean in vitro
`dissolution time to either the mean residence time or the
`mean in vivo dissolution time (77,78), so it does not reflect
`the actual in vivo plasma concentration curve. Therefore, level
`B correlation alone cannot support biowaivers. A level C
`correlation represents a single-point relationship between a
`dissolution parameter (e.g., t50%, t90%) and a pharmacokinetic
`parameter (e.g., AUC, Tmax, Cmax). This correlation does not
`reflect the entire plasma-concentration–time curve or dissolu-
`tion profile (62,79); therefore,
`it is considered the lowest
`correlation level. However, Level C correlation can provide
`useful information in early formulation development. For a class
`I drug, if the permeability is site-dependent, a level C correlation
`is expected (76). A multiple level C correlation compares one or
`more pharmacokinetic parameters of interest (e.g., Cmax, AUC)
`to the amount of drug dissolved at several time points of the
`dissolution profile. This level of correlation may support a
`biowaiver if the correlation has been established over the entire
`dissolution profile with one or more pharmacokinetic parame-
`ters of interest. If a multiple level C correlation is possible, then
`it is likely that a level A correlation is possible as well, and the
`latter is the preferred correlation. In addition, level D correla-
`tion, which is a qualitative a rank-order correlation, has been
`described in an FDA guidance (62). This correlation can be
`useful
`in drug development but cannot support regulatory
`application.
`Compared with immediate-release (IR) products (80,81),
`more attention has been given to the application of IVIVC
`for controlled-release oral dosage formulations (82–85),
`where formulation technology controls the release rate, thus
`drug release is the rate-limiting factor in the absorption
`process. For BCS class II drugs with immediate-release (IR)
`formulations, because of the intricacy of gastric emptying as
`
`
`
`1294
`
`Gray et al.
`
`well as the low resolution of plasma data at early time points
`(0–3 h), a meaningful level A correlation seems unlikely and
`few publications have been made so far. However, Lue et al.
`(86) and Buch et al. (87) recently applied biorelevant
`dissolution media (BDM) in the investigation of the IVIVC of
`class II, immediate-release (IR) compounds. The application of
`IVIVC to non-oral products, such as parenteral depots or
`injectable dosage forms, has also been investigated (72,88–93).
`It is worthwhile to mention that for IR drugs, in vitro–in vivo
`re