Dissolution and In Vitro- In Vivo Correlation

gxpand jv t .com Journal of Validation technology [Winter 2010] 57

Product and Process Design.Coordinated by Yihong Qiu]

For more Author

information,

go to

gxpandjvt.com/bios [

“Product and Process Design” discusses scientific and technical principles associated with pharmaceutical product development useful to practitioners in valida-tion and compliance. We intend this column to be a useful resource for daily work applications.

Reader comments, questions, and suggestions are needed to help us fulfill our objective for this column. Please send your comments and suggestions to col-umn coordinator Yihong Qiu at [email protected] or to journal coordinating editor Susan Haigney at [email protected].

KEY POINTSThe following key points are discussed in this article:

•Dissolution is an essential step in the drug absorption from solid products. In vitro dissolu-tion is primarily a function of drug solubility, the dosage form design, and the testing methodology and conditions

•There are different mathematical models that describe dissolution phenomena

•Various dissolution apparatus are adopted offi-cially by pharmacopeia. The United States Phar-macopeia (USP) Apparatus I and II are the most commonly used for solid oral dosage forms

•Different drug release mechanisms are employed in different drug delivery systems that can be described by various kinetic models

•Comparison of dissolution profiles are often per-formed using statistical approaches (e.g., f1 and f2)

•Physico-chemical property of the drug substance, dosage form design, medium, and apparatus are important considerations in developing a dis-solution method

•In vitro-in vivo correlation (IVIVC) refers to a quan-titative relationship between in vitro properties (usually dissolution) and the in vivo performance (e.g., drug plasma concentration-time profile) of a drug product

•IVIVC is categorized into level A, level B, level C, and multiple level C, with level A, level C, and multiple level C being the most useful for regulatory applications

•Considerations in establishing an IVIVC include development of a predictive in vitro test, in vivo study design, model building, and model validation

•A level A IVIVC may be established by a two-stage (deconvolution followed by correlation) approach or a single-stage (convolution followed by com-parison of drug-plasma profile) approach

•Internal or external validation of IVIVC models is required depending on specific situations

•When an IVIVC is established and validated, in vitro dissolution becomes a surrogate for the in vivo performance of the drug product. Therefore, it can be used to determine meaningful dissolution specifications, to obtain waivers for bio studies

Understanding Biopharmaceutics Properties for Pharmaceutical Product Development and Manufacturing II—

Dissolution and In Vitro-In Vivo CorrelationDeliang Zhou and Yihong Qiu

ABOUT THE AUTHORSDeliang Zhou, Ph.D., is a principal pharmaceutical scientist in Oral Drug Products, Manufacturing Science and Technology at Abbott Laboratories. He may be reached at [email protected] Qiu, Ph.D., is a research fellow and associate director in Global Pharmaceutical Regulatory Affairs CMC, Global Pharmaceutical R&D at Abbott Laboratories. He may be reached at [email protected].

58 Journal of Validation technology [Winter 2010] i v t home.com

Product and Process Design.

required for certain scale-up and post-approval changes (SUPAC)

•The success of IVIVC exploration depends on a multitude of factors including the physico-chemi-cal, biological, and pharmacokinetic properties of the drug molecule, the dosage form design, and their interplay in the gastrointestinal tracts. It remains a challenging task

•Validation personnel should be especially vigilant regarding changes or variations in materials, pro-cess, equipment, and scale that may impact dissolu-tion, particularly for modified-release products.

INTRODUCTIONThe therapeutic efficacy of a dosage form is determined by the drug concentration in the body or at the site(s) of action, which is dependent on the rate and extent of drug absorption and disposition in the body. In vivo dissolu-tion and the subsequent permeation of drug molecules across the intestinal membrane into the blood stream are two essential steps in oral drug absorption from a solid dosage form. While the permeation and disposi-tion are inherent properties of a particular drug, the in vivo dissolution depends greatly on the properties of the drug, dosage form design, and drug release environment in the gastrointestinal tract (GI) tract.

Over the last half century, in vitro dissolution has evolved as a key quality and performance measure of modern drug products. In general, dissolution of an oral product is influenced by the physicochemical proper-ties of the drug substance, the dosage form design, the manufacturing process, and the testing conditions (i.e., apparatus, agitation, medium, etc.). During the develop-ment of a drug product, a dissolution test is used as a tool to identify factors that influence or may have a potential effect on the oral absorption and to guide formulation and process selection. Once the composition and the process are defined, a dissolution test is utilized in pro-cess scale-up, optimization, validation, and in quality control of production batches to ensure batch-to-batch consistency. In certain instances, a dissolution test can be shown to differentiate between bioequivalent and non-bioequivalent batches, which is useful in demonstrating bioequivalence of drug products. The dissolution test may also be used to establish a quantitative relationship between the in vitro drug release and in vivo absorption (i.e., in vitro-in vivo correlation [IVIVC] when it is shown to predict the in vivo absorption characteristics of a drug product). IVIVC is an important tool of significant utility during formulation and process development, quality control, and many regulatory applications involving

changes to formulation, manufacturing process, equip-ment, manufacturing site, and other variables.

IN VITRO DISSOLUTIONDissolution is the process by which a solid enters a sol-vent phase to form a solution. In the pharmaceutical industry, in vitro dissolution is one of the most important physicochemical tests of drug products. It has a long and interesting history itself (1). Dissolution was not a regulatory requirement until after its association with drug product performance was studied in the 1950s and 1960s. In 1951 Edwards postulated the dissolution of aspirin tablet in the gastric environment would be the rate-limiting process of its oral absorption. Nelson established experimentally a relationship between in vitro dissolution and blood levels of theophylline in 1957. From the 1960s to early 1970s, the relationships between dissolution and clinical efficacy or the lack thereof, and in some cases the toxicity, were demonstrated for a num-ber of drug products, including tolbutamide, sodium diphenylhydantoin, chloramphenicol, sulfisoxazole, ampicillin, digoxin, and phenytoin (1). For example, it was found that simply replacing the filler calcium sulfate with lactose in phenytoin tablet resulted in toxicity due to much faster drug release and higher plasma concentration of phenytoin in a large number of patients (2).

With the increasing evidence of the association between dissolution and the drug product performance, the official dissolution test was finally adopted by the United States Pharmacopeia (USP) in 1970. Since then, a number of revisions and additions have taken place as a result of advances in research on dissolution, oral absorption, and bioavailability.

Theory Of DissolutionStudy of dissolution phenomenon can be traced back to as early as the late 19th century. Noyes and Whitney (3) first found the proportionality between dissolu-tion rate and the difference between solubility of the solutes, Cs, and the concentration in the bulk, C:

[Equation 1]

where k is a constant. Further experiments by Brunner and Tolloczko in 1900 established a relationship between rate of the dissolution and the exposed surface (4):


Coordinated by Yihong Qiu.

[Equation 2]

where S is the surface area. A few years later Brunner and Nernst eventually developed what is now known as the Nernst-Brunner dissolution equation (5):

[Equation 3]

Where D is the diffusion coefficient, V is the vol-ume of the dissolution media, and h is the thickness of the diffusion layer. This equation is based on the diffusion layer concept and Fick’s second law, which laid the foundation for dissolution modeling.

In studying dissolution of spherical particles, Hix-son and Crowell (6) derived the cubic-root law of dissolution, by applying the relationship between the surface area and the mass of spherical particles:

[Equation 4]

where M0 and Mt are the particle mass at time zero and at time t, respectively, r0 is the initial particle radius, and ρ is the density of the particles.

Other particle dissolution models also exist, such as the two-thirds root model by Higuchi and Hiestand, where the thickness of the diffusion layer is assumed to be sub-stantially greater than the particle radius.

Flanagan and Wang considered surface curvature during particle dissolution and developed a general particle dis-solution equation based on diffusion layer model (7):

[Equation 5]

This equation better describes particle dissolution under all conditions while other variations are approx-imations under specific conditions. For example, the cubic-root law only applies to the situation where dif-fusion layer thickness is much less than the particle size (i.e., the particles are large so that the surface curvature can be ignored). It is worth noting that the curvature of diffusion layer was correctly considered by Ozturk et al. (8) when concerning the dissolution of ionizable drugs.

It should be pointed out that the diffusion layer model is not the only plausible physical explanation of dissolution process. The macroscopic packets model by Danckwerts is a model that assumes the new sol-vent packets constantly reach the solid surface, absorb solute molecules, followed by delivery to the bulk solution. This model can be conveniently used to explain the influence of hydrodynamics (e.g., agita-tion) on dissolution. Levich (9) combined elements from both models and developed the following model that put both diffusion and hydrodynamics together into consideration:

[Equation 6]

Where f and f ’ are constant, υ is the solvent kine-matic viscosity, and V∞ is the bulk fluid velocity that may be replaced by rotation rate ω when the test appa-ratus is driven by a rotation mechanism.

Dissolution Test MethodThe dissolution test has several distinct components that include an apparatus, a dissolution medium, a sample assay method, and acceptance criteria. As with any analytical test, a dissolution method needs to be developed and validated to ensure its accuracy, reproducibility, and robustness.

A variety of designs of apparatus have been used for dissolution testing. Many of the tests with recom-mended apparatus and specifications are available in the pharmacopoeias. In the USP general chap-ter <711>, seven dissolution apparatus are currently defined. Apparatus I (rotating basket) and Apparatus II (rotating paddle) are the most commonly used for oral dosage forms. Apparatus III is a reciprocating cylinder, Apparatus IV is a flow-through cell, Appa-ratus V is paddle over disk, Apparatus VI is a rotating



cylinder, and Apparatus VII is a reciprocating holder. Of these, Apparatus V and VI are not designed for oral dosage forms. For example, paddle over disk as well as rotating cylinder are often used in the testing of transdermal products. Apparatus III uses a reciprocating cylinder to induce agitation and move dosage form through a series of vessels. It is convenient for monitoring release in dif-ferent test media such as different pH buffers. Apparatus VII is a similar design but with much smaller volume (10 versus 250 mL), which can be useful for testing drug release of small doses, such as drug-eluting stents.

Both Apparatus I and II use a cylindrical vessel with a hemispherical bottom. The hydrodynamics is provided by centered spindle rotation mechanism. A basket that houses the dosage form is attached to the end of the spindle for Apparatus I. A paddle is attached to the end of the spindle for Apparatus II and the dosage form is placed at the bottom of the vessel. A device to submerge the dosage form may be used with Apparatus II to prevent floating of dosage forms—often seen with hard gelatin capsules. The bottom of the basket or the paddle is set to a distance of 2.5 cm from the bottom of the vessel. The rotation speed for Apparatus I is typically 75-100 rpm while 50-75 rpm are often used for paddle method. Typical medium volume is in the range of 500-1000 mL, with 900 mL the most common volume. Samples are usually taken at the midway between the medium surface and the top of the basket/paddle blade, and not less than 10 mm from the vessel wall. To maintain the hydrodynamics and constant volume, sampling with medium replacement is preferred, particularly when volume of each sample is relatively large.

The performance of Apparatus I and II is influenced by a number of factors related to the instrument setup, such as centering and verticality of the shaft, the level-ing of the bath, and vibration. USP provides calibrator (prednisone and aspirin) tablets to ensure consistent performance across equipment and laboratories. The hydrodynamics of Apparatus II have been shown to be inhomogeneous by experiment and simulation. Particu-larly, a hydrodynamic “dead zone” has been found at the bottom of the USP vessel where dissolution takes place. Minor variations in the position of the paddle can cause variability in dissolution. A number of modifications to the current apparatus have been studied. Neverthe-less, challenges in obtaining more consistent dissolution results remain.

Dissolution testing is generally conducted at 37°C. High performance liquid chromatography (HPLC) is the preferred analytical method for sample assay, although the ultraviolet (UV) method may also be used.

Evaluation Of Drug Release DataDrug release is determined by the release mechanism of the dosage form, the physico-chemical proper-ties of the drug molecule (and the excipients), and the conditions of dissolution testing. For immediate release products, the rate and extent of drug release are often evaluated using the dissolved amounts (Q) at specified time points (t). When an adequate num-ber of data points are available, empirical or semi-empirical models have also been used to describe the dissolution behavior, such as, first order, cube-root, and Weibull models. The following Weibull function is an example of using an empirical equation used to describe dissolution (10):

[Equation 7]

where tlag is the lag time, α is a scale constant, and ß is a shape constant. Weibull function offers flex-ibility as it can describe exponential release (ß =1), sigmoid profile retarded in the beginning (ß >1), and sigmoid profile retarded in the end (ß <1).

For extended release (ER) dosage forms, many mod-els have been developed based on considerations of the principles of product design. The following dis-cusses examples of drug release models for a number of common ER drug delivery systems.

When a soluble drug is uniformly dispersed in an insoluble matrix, drug release is usually controlled by diffusion via leaching in the dissolution medium. The distance of drug diffusion increases with time in this case. The following is a well-known square-root-of-time release model for a finite planar system derived by Higuchi (11):

[Equation 8]

Where Q(t) is the amount of drug release at time t, A is the loaded drug concentration in the matrix, and k is a constant. It should be noted that there are mathematically more accurate equations that describe drug release from finite planar systems via



the diffusion mechanism; the Higuchi equation is an approximation valid up to about 65% of release; and in a three-dimensional system, the surface area available for drug release also decreases with time.

A reservoir delivery system consists of a drug-contain-ing core enclosed by a permeable membrane. The driv-ing force for drug release is the concentration gradient across the rate-controlling membrane. When the drug loading is sufficiently high, the concentration gradient is almost constant, thus providing zero-order drug release. As drug depletes from the core, the concentration inside the membrane will eventually fall below the solubility, resulting in deviation from the zero-order kinetics in the later part of the drug release. Multi-particular res-ervoir system is the most common. These are usually drug-layered inert spherical carriers (e.g., sugar or cel-lulose spheres) coated with a water insoluble membrane (e.g., ethylcellulose) containing pore formers. The pore formers are usually hydrophilic component (e.g., poly-ethylene glycols [PEG], hydroxypropyl methylcellulose [HPMC]). The advantage with multi-particulate system includes minimized risk of dose dumping and flexibility in adjusting dose or strength.

The osmotic pump is another type of drug delivery system that provides zero-order drug release. It is similar to a reservoir device but contains an osmotic agent that acts to imbibe water from the surrounding medium via a semi-permeable membrane. Such a device, called the elementary osmotic pump (EOP), was first described by Theeuwes and Higuchi (12) in 1975. The release of the drug from the device is controlled by water influx across the semipermeable membrane. The drug is forced out of an orifice in the device by the osmotic pressure generated within the device. The size of the orifice is designed to minimize solute diffusion, while preventing the build-up of a hydrostatic pressure head that has the effect of decreasing the osmotic pressure and changing the volume of the device. The rate of drug release is proportional to the difference between the osmotic pressure, Δπ, and the hydrostatic pressure difference, ΔP:

[Equation 9]

Where k is a constant, S is the surface area of the membrane, h is membrane thickness, and C is the drug concentration inside the membrane. When ΔP

is negligible and there is an excess of solid (saturated solution) in the device, the drug release rate remains constant delivering a volume equal to the volume of water uptake. In developing oral products, two types of osmotic pump systems have been frequently used, a one-chamber EOP system, and a two-chamber system (e.g., Push-Pull).

A swellable hydrophilic polymeric matrix system is most commonly used for developing oral extended release products. The release mechanism from the matrix system is often a combination of drug diffusion and polymer erosion. In the presence of water, the hydrophilic polymers undergo swelling, formation of a gel layer and subsequent polymer dissolution (erosion), which controls the rate of drug release. Insoluble drugs tend to impart drug release by poly-mer erosion mechanism, while release of soluble drugs is primarily controlled by diffusion because the time scale of diffusion is less than the polymer erosion. No exact mathematical model exists that is capable of describing drug release kinetics from this type of system with concurrent swelling, diffusion, and ero-sion. In the mid-1980s, Peppas (13) introduced a semi-empirical power law equation that has been widely used to describe drug release behavior from hydrophilic matrix systems:

[Equation 10]

Where k is the rate constant and n is the release exponent. It has been shown that the value of n is indicative of the drug release mechanism. For n = 0.5, drug release follows a Fickian diffusion mechanism that is driven by a chemical potential gradient. For n = 1, drug release is primarily controlled by polymer erosion. For 1 > n > 0.5, non-Fickian diffusion behav-ior is often observed as a result of contributions from diffusion and polymer erosion.

Comparison Of Dissolution ProfilesComparing dissolution performance plays an impor-tant role in the previously-mentioned changes asso-ciated with composition, manufacturing process, equipment, and site, etc. Under appropriate test conditions, a dissolution profile can characterize a product more precisely than a single point dissolu-tion test. Thus, a dissolution profile comparison between pre-change and post-change products or



different strengths helps assure similarity in product performance. General approaches that are used to com-pare the entire dissolution profiles include analysis of variation (ANOVA)-based, model-independent, and model-dependent methods.

In a model-dependent approach, the dissolution data are first fit to an appropriate drug release model. Quantitative evaluation of the fitted parameters or rate constant is then used to compare the dissolution profiles. However, defining a common model that accurately and reliably describes different profiles is often very challenging.

Model-independent approaches are more frequently used, which usually adopt statistical approaches to evaluate directly the differences between a test profile and a reference profile. The most commonly used are the difference factor, f1, and the similarity factor, f2, as defined in the following equations (14):

[Equation 11]

[Equation 12]

Where n is the number of time points in the dissolu-tion profile, Rt is the mean release from the reference product at time t, while Tt is the mean release from the test product at time t. The difference factor calculates the mean absolute differences between two dissolution profiles. Two perfectly identical profiles have f1 of zero, and a value of less than 15 is often interpreted as no difference. The similarity factor is based on the sum of square of difference in release. Two identical profiles lead to f2 of 100 and a 10% difference at each time point results in an f2 of 50. Therefore, an f2 value of 50-100 is used to demonstrate the sameness between two dis-solution profiles. The drawback in using either f1 or f2 is that the results are often sensitive to the number of time points and how the time points are selected. Limiting to only one time point at >85% release is recommended.

General Considerations In Dissolution TestingDrug dissolution profile is influenced by a multitude of factors including physico-chemical properties of the drug substance, the formulation design, the manu-facturing process, and the physical, chemical, and mechanical condition of the test.

Physico-chemical Properties. Among various properties of the active pharmaceutical ingredient (API), solubility is one of the most important factors that govern drug release from solid oral dosage forms. For example, the rate of particle dissolution is directly proportional to the solubility in the testing medium. Drug solubility plays a key role in the release rate of the reservoir deliv-ery system. When testing poorly water-soluble drugs, solubilizers such as surfactants are often incorporated in the dissolution medium to improve solubility and create sink conditions. Solubility change as a result of changing solid-state forms of the drug substance (e.g., polymorph, hydrate/solvate, and salt forms) is known to affect the release rate or mechanism. In some instances, drug release may be impacted by physical and chemical interactions between drug and excipients or between excipients. Another important property that should be considered in dissolution test method development is chemical stability.

Dissolution Medium. Dissolution testing of oral dosage forms is often conducted in aqueous media of gastric pH (pH 1-2) and/or intestinal pH (e.g., pH 6.8, pH 7.4). Dissolution at intermediate pH (e.g., pH 4.5) may also be necessary in certain instances. Many drug molecules are weak acids or weak bases. Thus, important factors to consider in the selection of a test medium include solubility and its depen-dency on pH of the drug substance, as well as pH and buffering capacity of the test medium. As discussed previously, solubilizers or surfactants may be added to a dissolution medium to achieve sink conditions. Testing in multiple dissolution media (pH 1 to 6.8) is often necessary in justifying certain scale-up and post-approval changes (SUPAC). Pure water is not preferred due to its lack of buffering capacity as the dissolved drug may significantly alter the medium pH during testing.

Dosage Form Design. Dissolution characteris-tics are defined by the objectives of product design. Immediate-release (IR) dosage forms are designed to render drug molecules available for absorption imme-diately following administration. In most cases, they are required to reach complete release within a short period of time (e.g., 30-60 min). Extended release



dosage forms are intended to gradually release the drug over an extended period of time. The dissolution profile of an ER dosage form is needed for evaluating the release characteristics. To control product quality, at least three time points are required: the first point is used to demonstrate absence of dose-dumping, the last point is to ensure complete release (>80%), and the middle points are intended to measure proper control of the release rate. Enteric coated tablets are intended to prevent drug release in the stomach (acidic environment) based on clinical needs (e.g., the gastric irritation) or concerns of acid instability, or other delivery considerations. The dissolution method of enteric coated products usually uses a two-stage approach: An acid stage at pH 1 to demonstrate the integrity of the enteric coating, and a second stage at intestinal pH for drug release. The amount of drug release in the acid stage is usually limited to less than 10%. After the acid stage, drug release from enteric products can be immediate or controlled, depending on the specific design objective.

Acceptance Criteria. Acceptance criteria of dis-solution for quality control are determined based on product design, dissolution characteristics of batches used in pivotal clinical or bioavailability studies, and consideration of regulatory requirements, compendial standards, and analytical test methods.

The acceptance criterion for an IR dosage form is usually defined by a minimum amount of drug release, the Q value, at a single time point (e.g., Q ≥80% release at 30 minutes). Stage testing is normally allowed based on a combination of the results from individual tablet and the mean release. For example, USP considers the drug release criterion met if the release from the first six tablets ≥ Q + 5% (Stage 1). Otherwise, an additional six tablets are tested (Stage 2). To meet the Stage 2 criterion, the combined results should have a mean dissolution ≥ Q and no individual unit is less than Q–15%. If Stage 2 fails, additional 12 units are tested (Stage 3). In order to pass Stage 3, the combined results should satisfy: (a) mean ≥ Q; (b) not more than 2 units release less than Q–15%, and (c) no unit release less than Q–25%. No testing is permissible beyond Stage 3.

The dissolution specifications for ER dosage forms are set based on the dissolution data from pivotal bio-batches. Where IVIVC is absent, the allowable range of the dissolution limits is ±10% label amount at each time point from the mean of the biobatch. Similar to stage testing, level testing may be permitted for ER dosage forms depending on the pharmacopeia.

For example, USP allows Level 2 and Level 3 testing when failure of the previous level exists. In each level testing, the acceptance criterion for an individual dose unit is widened by ±10% label amount from the previous level. The acceptance criterion for the mean is always within the Level 1 specification limits at each time point. When IVIVC is established, Level 1 dissolution specification limits wider than ±10% label amount may be justified.

IN VITRO-IN VIVO CORRELATIONAn in vitro-in vivo correlation of a drug product can be established when an in vitro dissolution test is shown to be predictive of in vivo absorption. Since the early 1990s, IVIVC has received considerable attention from the industry, regulatory agencies and academia, par-ticularly since the publication of the guidance docu-ments on dissolution testing of immediate release and IVIVC of extended release dosage forms by the US Food and Drug Administration and European regula-tory authorities (EMEA) in the mid- to late-1990s. As a result, there has been increased investigation and success in using in vitro tests to evaluate or predict in vivo performance of solid drug product, especially ER dosage forms. With an established IVIVC, the dissolution data can be used not only as a quality control tool, but also for guiding and optimizing prod-uct development, setting meaningful specifications, and serving as a surrogate for bioavailability studies required for certain types of product and manufactur-ing changes or variations.

USP defines IVIVC as the establishment of a rela-tionship between a biological property, or a parameter derived from a biological property produced by a dos-age form, and a physicochemical characteristic of the same dosage form. FDA defines IVIVC as “A predic-tive mathematical model describing the relationship between an in vitro property of a dosage form (usually the rate or extent of drug dissolution or release) and a relevant in vivo response (e.g., plasma drug concentra-tion or amount of drug absorbed).” (15)

Categories Of IVIVCCurrently, IVIVC is categorized into level A, level B, level C, and multiple level C by the regulatory agen-cies, as follows:

•Level A correlation represents a point-to-point rela-tionship between the in vitro dissolution and the in vivo input rate (e.g., the in vivo dissolution from the dosage form) or the in vivo response time course (e.g., plasma concentration).



•Level B correlates the summary parameters that char-acterize the in vitro and in vivo time courses based on statistical moment analysis (e.g., correlations between the mean in vitro dissolution time to the mean in vivo dissolution time, or to mean in vivo residence time). This is not considered as a point-to-point mapping because a number of profiles may produce similar mean statistical parameters.

•Level C correlation establishes a single point relation-ship between a dissolution parameter, for example, amount dissolved in vitro at a particular time point (e.g., Q60) or the time required for dissolution of a fixed amount (e.g., t50%), and a summary parameter that characterizes the in vivo time course (e.g., AUC, Cmax, or Tmax).

•A multiple Level C correlation establishes the rela-tionship between in vitro dissolution and one or several pharmacokinetic parameters of interest, at multiple time points.

In supporting regulatory applications and product development, Level A correlation is the most informa-tive and useful as it can be used to predict the entire in vivo plasma profile from the in vitro release data. Its key applications include ensuring product quality by setting meaningful dissolution specifications, and justifying waiver of bioequivalence studies (biowaivers) required for various post-approval changes. Level C correlation can provide useful guidance in product development. Multiple Level C correlation can also be used to help setting dissolution specification or support biowaivers in many cases. Level B correlation is least useful for regulatory applications because different in vitro or in vivo profiles may have similar mean time values.

Convolution And DecovolutionConvolution can be found in many applications of phys-ics, engineering, and mathematics wherever there is a linear system. Computing the inverse of the convolu-tion operation is known as deconvolution. Convolution and deconvolution techniques play an important role in understanding and establishing a Level A IVIVC. A linear system is a system based on the use of a linear operator. In simple words, the behavior of a linear system is addi-tive (superposition principle applicable) (i.e., it can be predicted by summing up all of its historical events). This mathematical property makes the solution of modeling equations simpler than those of nonlinear systems.

The body’s response to a drug input can be regarded as a linear system in most cases. Nonlinear response is more complex and is beyond the scope of this discus-

sion. In developing IVIVC, one of the critical steps is to understand and obtain in vivo drug input for correlation with the in vitro drug release from the dosage forms. In general, the plasma concentration-time profile after a unit dose of intravenous (iv) bolus contains all informa-tion with regard to drug disposition in the body. Thus, it is defined as a unit impulse function, g(t) (i.e., it is characteristic of the body system). When a drug mol-ecule is introduced into the systemic circulation at a rate described by a non-instantaneous input function (e.g., in vivo drug release), f(t), the system response (e.g., plasma concentration resulting from the in vivo drug release), r(t), can be calculated via convolution of f(t), and g(t),:

[Equation 13]

or, in simpler notion:

[Equation 14]

The unit impulse function is also known as the system function, the weighting function, the characteristic func-tion, the disposition function, or the kernel. In IVIVC application, the unit impulse function is not limited to the body response after iv bolus dose. It can also be esti-mated using plasma concentration following administra-tion of solution or immediate release (IR) dosage forms depending on the situations given in the Table.

In pharmacokinetic studies, the responses (e.g., plasma concentrations or urinary secretion) after a unit impulse (a reference dose), g(t), and from a non-instantaneous input, r(t), are usually measured experimentally while the in vivo input function, f(t), is the unknown function. In a linear system, f(t) can be calculated using deconvolution which is the inverse operation of convolution. Alterna-tively, deconvolution can also be applied to estimate the system function, g(t) when f(t) and r(t) are known. Decon-volution may be written as the following notion:

or

[Equation 15]



A detailed discussion on how to perform a convolution or deconvolution (16) is beyond the scope of this column. In summary, the objective of convolution application in IVIVC is to calculate the plasma drug concentration-time profile of a product given its input rate and the plasma drug concentration profile of a reference dose (impulse function). Deconvolution is utilized to estimate the input rate of a formulation given its plasma level-time profile and the impulse function. Various mathematical methodologies are available to perform such tasks. Many commercial software packages have been developed for this purpose.

Development And Assessment Of IVIVCThe key components in establishing an IVIVC include study design, model building, and model validation. In its 1997 guidance (17), FDA presented a compre-hensive perspective on methods of developing and evaluating an IVIVC, setting dissolution specifications using an IVIVC, and using dissolution as a surrogate for bioequivalence study required for certain scale-up and post-approval changes (SUPAC).

Study Design. To demonstrate an IVIVC, formu-lations with different dissolution rates and in vivo per-formance are needed. Usually two or more (preferably three or more) formulations are used. Under certain circumstances, use of only one formulation may be possible, when the in vitro release profile is shown to be independent of the dissolution test conditions.

All formulations should be tested under the same dissolution conditions and the dissolution profiles should differ sufficiently (e.g., by > 10%) to allow for a meaningful correlation. The in vitro data are often generated in an aqueous medium using USP Apparatus I or II, operating within an appropriate range of rotation speeds (e.g., 50–100 rpm). Other methodologies may be used when justified. A mini-mum of 12 individual dosage units should be tested, with coefficient of variation less than 10%. The key is to develop a dissolution methodology to adequately discriminate among formulations. The dissolution methods may need to be revisited based on the in vivo

performance of the formulations, as demonstrated in the case of divalproex sodium (18-20).

Sufficient human subjects should be included in the bioavailability study, in order to adequately char-acterize the absorption profiles of the drug products. The number of subjects may depend on the inherent variability of a particular drug molecule in humans. The guidance prefers crossover studies, but parallel studies or cross-study analyses are also acceptable, provided there is a common reference (e.g., iv, oral solution, or IR dosage form) treatment. In addition, fasted state is preferred. When a drug is not toler-ated in the fasted state, studies may be conducted in the fed state.

IVIVC Model Building. The following discusses IVIVC model building:

•Level A correlation. Level A correlation is often developed using either a two-stage or a single-stage procedure. The two-stage approach is the most frequently used in building IVIVC models. It involves estimation of the in vivo release and absorp-tion profile from plasma concentration-time data using an appropriate deconvolution technique for each formulation, followed by comparison with the percent released in vitro, preferably using a lin-ear model. A slope close to unity indicates a 1:1 correlation between in vitro dissolution and in vivo absorption/dissolution. A negative intercept implies that the in vivo process lags behind the in vitro dissolution. A positive intercept has no clear physiological meaning. It can be a result of relatively high variability or curvature at the early time points. A scaling factor, such as time-shifting and time-scaling parameters, may be used when necessary. Nonlinear models, while uncommon, may also be appropriate. The figure provides a typical model building process based on two-stage approaches. Alternatively, Level A IVIVC can be estimated with a single-stage approach, where the time course of plasma concentrations is directly predicted by con-voluting the dissolution data based on a projected

Table: Typical system definitions in linear system analysis for oral delivery.

CaseUnit impulse response g(t)

Input response r(t)

Input function f(t)

I Plasma levels from iv bolus Plasma levels from oral solution Absorption in GI

II Plasma levels from iv bolus Plasma levels from oral solid dosage form Dissolution and absorption in GI

III Plasma levels from oral solution (or IR dosage form)

Plasma levels from oral solid dosage form Dissolution in GI



model. The plasma concentration profiles pre-dicted from the model are then compared directly to the observed ones. This approach has been illustrated with carbamazepine (21).

•Level B correlation. Level B IVIVC compares mean times that molecules spend in in vitro and in vivo spaces. It requires in vivo testing of at least three dosage forms. Based on statistical moment analysis (22), mean residence time (MRT), mean absorption time (MAT), or mean in vivo dissolu-tion time (MDTin vivo) can be calculated and related to mean in vitro dissolution time (MDTin vitro). The analysis is very general without assumption on kinetics of the drug transfer in a system. All parameters are model-independent. The corre-lations are reflective of the entire dissolution or plasma concentration curves by using all available in vitro and in vivo data. One major issue is that curves of different shapes may produce similar or same mean time values.

•Level C correlation. Level C correlation requires in vivo testing of three or more formulations having different release rates for establishing a linear or nonlinear relationship between in vitro and in vivo parameters as each data point of the correlation plot corresponds to one formulation. The commonly used summary parameters for in vivo profile are AUC, Cmax, Tmax, while the in vitro parameters are often percent released at time t (Qt). A single point Level C correlation may be helpful for formulation development. It can also be used to justify setting dissolution specification at a particular time point. The information is generally insufficient for justi-fying a biowaiver. A multiple Level C correlation establishes correlation at several time points with one or more bioavailability parameters of interest. If the correlation covers the entire dissolution profile, it may be used to support a biowaiver. A multiple Level C correlation should be based on at least three dissolution time points covering the early, middle, and late (preferably > 80% release) stages of the dis-solution profile. The establishment of a multiple Level C correlation often warrants further explora-tion of Level A correlation.

Model Validation. A single IVIVC equation should hold for all formulations investigated. If a formulation with the highest or lowest release rate deviates signifi-cantly from the relationship, the correlation may still be valid over the range of release rates defined by rest of the formulations.

The FDA guidance laid out detailed criteria to assess the predictive performance of a proposed IVIVC through the estimation of prediction error (PE). It is defined as: %PE = [(observed value – predicted value) / observed value] × 100. Determination of PE involves calculation of the in vivo absorption profiles from the in vitro data using the established IVIVC model, followed by prediction of the corresponding plasma concentration profiles via convolution. The guidance further elaborates approaches to validate the model internally and externally.

The internal PE evaluates how tightly the model describes the data used to develop the IVIVC. The inter-nal validation may suffice when the IVIVC are derived using three or more formulations with different release rates for a non-narrow therapeutic index drug. Internal validation requires that the mean absolute %PE should be within 10% for Cmax and AUC, and that the %PE for each individual formulation should not exceed 15%.

The external validation requires a data set that was not used in the original development study of the IVIVC. The test product may include formulations with different release rates, formulations with minor manufacturing process changes, or less preferable, formulations from a different manufacturing batch. The external validation affords greater “confidence” in the model, and is recom-mended by FDA for the following scenarios:

•Narrow therapeutic index drug•Only two formulations were used to develop

the correlation•Calculation of the internal PE is inconclusive.

A limit of 10% PE is also required to establish an external validation. PE of 10% to 20% indicates inconclusive predictability and illustrates the need for further study using additional data sets. PE of greater than 20% indicates lack of predictability.

Multiple Level C correlations should be based on at least three dissolution time points covering the early, middle, and late stages of the dissolution profile. The assessment of the predictability will depend on the type of application for which the correlation is to be used. The same methodology and criteria as in the case of Level A correlation are expected.

Applications Of IVIVCThe objective of developing an IVIVC is to identify an in vitro test that can be used to predict in vivo perfor-mance of drug products. When this is achieved, IVIVC can be used as a surrogate for human bioavailability studies, which is useful to set meaningful dissolu-tion specifications to ensure product quality, and to



obtain biowaivers for supporting certain formulation or manufacturing changes.

Setting Dissolution Specifications. As dis-cussed previously, dissolution specifications are derived from the bioavailability batches. For extend-ed-release dosage forms, the specification range at any time point has to be within ±10% of the mean profile of the pivotal bioavailability batch in the absence of an IVIVC.

A validated Level A correlation allows for setting dissolution specifications such that all lots within the specification limits will be bioequivalent. The proce-dure involves predicting the in vivo drug plasma-time course from dissolution profiles via convolution based on the IVIVC model. The dissolution specification is determined such that the dissolution profiles defined by the upper and lower limits will produce in vivo pro-files with a maximum difference of 20% in predicted Cmax and AUC, thus rendering a biological meaning to the specification limits and to the controls carried out with the in vitro dissolution test. In addition, a more flexible and wider dissolution specification (e.g., > ±10%) may be possible so long as the predicted Cmax

and AUC are within 20%.Similarly, in the case of Level C and multiple Level

C IVIVC models, specification ranges can be set at the correlation time point such that there is a maximum of 20% difference in the predicted AUC or Cmax. If the correlation involves more than one parameter, the one resulting in tighter limits should always be used. In addition, drug release at the last time point should be at least 80%.

Waiver of In Vivo Bioavailability Studies. A validated Level A or multiple Level C correlation can be used to justify waiver of bioequivalence stud-ies for the many scenarios where an in vivo study is required. Examples include post-approval changes of manufacture sites for ER products, certain levels of process, formulation, and equipment changes. For immediate release solid dosage forms, the biowaiver request for investigational new drug (IND) applica-tions, new drug applications (NDAs), abbreviated new drug applications (ANDAs), and post-approval changes should be based on the consideration of the drug’s Biopharmaceutics Classification System (BCS) class, therapeutic index (narrow or non-narrow), and

In vivo pro�les

Deconvolution

Model buildingTime (h)

0 4 8 12 16 20 24 28 320

0.5

1.0

1.5

2.0

2.5

3.0

ABC

C (µ

g/m

l)

In vivo dissolution

Time (h)

00

20

40

60

80

100

120

4 8 12 16 20

ABC%

Rel

ease

d

IVIVC model

% In vitro release0 20 40 60 80 100

0

20

40

60

80

100

ABC%

In v

itro

abso

rptio

n

In vitro drug release

Time (h)0

0

20

40

60

80

100

4 8 12 16 20

ABC

% R

elea

sed

Figure: Illustration of the two-stage approach.



potential effect of excipients on bioavailability. Com-prehensive regulation, scientific rationales, consider-ations, approaches and evaluation criteria associated with these biowaivers have been clearly laid out in various regulatory guidelines (16, 17, 23, 24).

ChallengesIVIVC development is often challenging because of the complex nature of drug absorption from the dos-age forms in the GI tract. Various factors that usually determine the feasibility and success of IVIVC include physicochemical, biological, and pharmacokinetic properties of a drug substance, the formulation design, in vitro and in vivo study design, modeling method-ology, and level of understanding of the interplays among many variables involved in the in vivo release and absorption. It should be noted that IVIVC is not expected when in vivo dissolution is not the rate-limit-ing step in drug absorption, such as in the case of IR dosage forms of BCS class I and III drugs. Dissolution often plays a critical role in the absorption of poorly soluble compounds. However, IVIVC of BCS class II or IV drugs from IR dosage forms remains a dif-ficult task. Compared with an IR product, an IVIVC is generally more likely for ER dosage forms where drug absorption is normally limited by drug release. As a result, IVIVC developments for ER dosage forms have been more frequently reported. However, the overall rate of success remains relatively low with the approved products.

IMPLICATIONS FOR VALIDATION AND COMPLIANCEThe information discussed in this article is funda-mental to the rational design and development of an oral drug product, its manufacturing process, release specifications, and testing prior to commercial distri-bution. Validation and compliance personnel should have a general understanding of the properties of the drug and drug products for which they are responsible. Specifically, they should be very aware of the drugs whose properties present high risks to manufacturing processes as well as processes that may potentially impact drug properties and product quality attributes such as solubility and dissolution. Modified release products require extra vigilance in this regard.

Dissolution test methods and specifications and IVIVC for the dosage form are determined as part of drug product development. This information and sup-porting documentation should be readily available to validation and compliance personnel. This informa-

tion should be useful in the evaluation and support of formulation and manufacturing changes.

Validation and compliance personnel should be especially vigilant of formulation and manufactur-ing changes that may impact product dissolution and IVIVC. Validation protocols developed in response to such changes should require appropriate sampling and testing in support of the changes. Knowledge of the dosage form complexity must also be considered in determining appropriate validation testing, especially when the IVIVC has been determined.

Manufacturing process changes that have potential to affect drug solubility or rate of dissolution are most likely to impact drug bioavailability. BCS class 2 and class 4 drugs, both of which characterize low solubility drugs, are most susceptible to these changes. Valida-tion and compliance personnel should be especially watchful to changes in granulation, drying, particle size reduction, and other processes with potential to impact the granule structure, granule or particle den-sity, particle size, solid form of drugs and excipients, or other physical properties known to affect dissolution and solubility. Less subtle changes, such as changes in the source of key excipients, should also be care-fully monitored. Consultation between development scientists and validation and compliance personnel are encouraged to thoroughly evaluate proposed changes in formulation and manufacturing processes.

Laboratory personnel must be particularly vigilant when testing high risk drugs and dosage forms. Dis-solution testing equipment and associated analytical instrumentation must be properly qualified and main-tained. Laboratory personnel must be appropriately trained, especially when testing sensitive products. Subtle dissolution apparatus effects may cause unex-pected test results for susceptible products. Testing of extended release products in which test procedures may extend to 24 hours require good laboratory per-formance by testing personnel.

SUMMARYDissolution test of drug products is a key performance test and quality control tool in the pharmaceutical industry. The dissolution performance of a drug prod-uct depends on the properties of the drug substance as well as the inactive excipients, the dosage form design, manufacturing process, and test conditions. A validated in vitro dissolution test can serve for the purposes of providing necessary quality and process control, determining stability of the relevant release characteristics of the product, and supporting certain



regulatory determinations and judgments concerning minor post-approval changes, etc. However, the dis-solution rate of a specific dosage form is essentially an arbitrary parameter that may vary with the test methodology and condition. Unless it is demon-strated that the in vitro release behavior reflects the in vivo performance in humans, the data can be of little relevant value in evaluating in vivo performance of a drug product. When an IVIVC is established, dissolu-tion becomes a surrogate for the in vivo performance of the drug product. Therefore, meaningful dissolution specifications can be developed for a more reliable quality control. Many other scientific and regulatory benefits of IVIVC are also possible. As far as valida-tion and compliance personnel are concerned, it is beneficial to understand the general aspects of dis-solution performance and how it may be related to various quality and manufacturing aspects of both API and drug product. Any changes or variations in materials, process, equipment, and scale require consideration of their possible impacts on dissolution, particularly for modified-release products.

REFERENCES1. Dokoumetzidis, A. and Macheras, P., “A Century Of

Dissolution Research: From Noyes and Whitney to the Biopharmaceutics Classification System,” Int. J. Pharm., 321, 1-11, 2006.

2. Tyrer, J. H.; Eadie, M. J.; Sutherland, J. M. and Hooper, W. D., “Outbreak of Anticonvulsant Intoxication in an Australian City,” Br Med J, 4, 271-3, 1970.

3. Noyes, A. and Whitney, W. R., “The Rate of Solution of Solid Substances in Their Own Solutions,” J. Am. Chem. Soc., 19, 930-934, 1897.

4. Bruner, L. and Tolloczko, S., “On the velocity of solu-tion of solid bodies,” Z. Phys. Chem. (Munich) 1900, 35, 283-290.

5. Nernst, W., “Theorie der reaktionsgeschwindigkeit in heterogenen systemen,“ Z. Phys. Chem. (Munich), 47, 52-55, 1904.

6. Hixson, A. W. and Crowell, J. H., “Dependence of reac-tion velocity upon surface and agitation. I. Theoretical considerations,” J. Ind. Eng. Chem. (Washington, D. C.), 23, 923-31, 1931.

7. Wang, J. and Flanagan, D. R., “General Solution for Dif-fusion-Controlled Dissolution of Spherical Particles. 1. Theory,” J. Pharm. Sci., 88, 731-738, 1999.

8. Ozturk, S. S.; Palsson, B. O. and Dressman, J. B., “Disso-lution Of Ionizable Drugs In Buffered And Unbuffered Solutions,” Pharm. Res., 5, 272-282, 1988.

9. Levich, V., Physicochemical Hydrodynamics, Prentice-Hall: Englewood Cliffs, NY, 1962.

10. Langenbucher, F., “Linearization Of Dissolution Rate Curves By The Weibull Distribution,” J. Pharm. Pharma-col., 24, 979-81, 1972.

11. Higuchi, T., “Rate Of Release Of Medicaments From Ointment Bases Containing Drugs In Suspension,” J. Pharm. Sci., 50, 874-5, 1961.

12. Theeuwes, F. and Higuchi, T., Osmotic dispensing de-vice with maximum and minimum sizes for the pas-sageway, US Patent 3916899, 1975.

13. Peppas, N. A., “Analysis Of Fickian And Non-Fickian Drug Release From Polymers,” Pharm. Acta Helv., 60, 110-11, 1985.

14. Moore, J. W. and Flanner, H. H., “Mathematical Com-parison Of Dissolution Profiles,” Pharmaceutical Tech-nology, 20, 64-74, 1996.

15. Tabusso, G., “Regulatory Aspects of Development Phar-maceutics (2),” Regulatory Affairs J., 12: 909-912, 1992.

16. FDA, Guidance for Industry: Dissolution Testing of Im-mediate Release Solid Oral Dosage Forms. U.S. Depart-ment of Health, Food and Drug Administration, Center for Drug Evaluation and Research. August 1997.

17. FDA, Guidance for Industry: Extended Release Oral Dosage Forms: Development, Evaluation, and Appli-cation of In vitro/In vivo Correlations. U.S. Department of Health, Food and Drug Administration, Center for Drug Evaluation and Research, September 1997.

18. Qiu, Y.; Garren, J.; Samara, E.; Cao, G.; Abraham, C.; Cheskin, H. S. and Engh, K. R., “Once-a-day controlled-release dosage form of divalproex sodium II: Develop-ment of a predictive in vitro drug release method,” J. Pharm. Sci., 92, 2317-2325, 2003.

19. Qiu, Y.; Cheskin, H. S.; Engh, K. R. and Poska, R. P., “Once-a-day controlled-release dosage form of dival-proex sodium I: Formulation design and in vitro/in vivo investigations,” J. Pharm. Sci., 92, 1166-1173, 2003.

20. Dutta, S.; Qiu, Y.; Samara, E.; Cao, G. and Granneman, G. R., :Once-a-day extended-release dosage form of di-valproex sodium III: Development and validation of a level A in vitro-in vivo correlation (IVIVC),” J. Pharm. Sci., 94, 1949-1956, 2005.

21. Veng-Pedersen, P.; Gobburu, J. V. S.; Mayer, M. C. and Straughn, A. B., “Carbamazepine level-A in vivo-in vitro correlation (IVIVC): A scaled convolution based predictive approach,” Biopharm. Drug Dispos., 21, 1-6, 2000.

22. Veng-Pedersen, P., “Mean time parameters in phar-macokinetics, definition, computation and clinical implications, Part I,” Clin. Pharmacokinet, 1989, 17, 345-366.



23. FDA, Guidance for Industry: Waiver of In Vivo Bio-availability and Bioequivalence Studies for Immedi-ate-Release Solid Oral Dosage Forms Based on a Bio-pharmaceutics Classification System. U.S. Department of Health, Food and Drug Administration, Center for Drug Evaluation and Research, August 2000.

24. FDA, Guidance for Industry: SUPAC-MR: Modified Release Solid Oral Dosage Forms Scale-Up and Post-approval Changes: Chemistry, Manufacturing, and Controls; In Vitro Dissolution Testing and In Vivo Bioequivalence Documentation, U.S. Department of Health, Food and Drug Administration, Center for Drug Evaluation and Research, September 1997. JVT

ARTICLE ACRONYM LISTINGANDAs Abbreviated New Drug ApplicationsANOVA Analysis of VariationAPI Active Pharmaceutical IngredientBCS Biopharmaceutics Classification System

EMEA European Medicines AgencyEOP Elementary Osmotic PumpER Extended ReleaseFDA US Food and Drug AdministrationGI Gastrointestinal TractHPLC High Performance Liquid ChromatographyHPMC Hydroxypropyl MethylcelluloseIND Investigational New Drug ApplicationIR Immediate ReleaseIVIVC In Vitro-In Vivo CorrelationMAT Mean Absorption TimeMDTin vivo Mean In Vivo Dissolution Time MRT Mean Residence TimeNDAs New Drug ApplicationsPE Prediction ErrorPEG Polyethylene ClycolsSUPAC Scale Up and Post-Approval ChangesUSP United States PharmacopeiaUV Ultraviolet

Dissolution and In Vitro- In Vivo Correlation

Documents

Transcript of Dissolution and In Vitro- In Vivo Correlation