For immediate-release dosage forms the in vitro dissolution process typically requires no more than 60 min, and in most cases a single time-point specification is adequate for Pharmacopeial purposes. To allow for typical disintegration times, test times of less than 30 min should be based on demonstrated need.

For extended-release products in vivo dissolution generally is rate limiting, which results in protracted drug absorption and thus facilitates the identification of in vitro test conditions that may be predictive of in vivo dissolution. Multiple sampling time points, therefore, are necessary to define a dissolution profile for a modified-release dosage form.

The choice of apparatus should be based on knowledge of the formulation and actual dosage form performance in the in vitro test system. Apparatus 1 (basket) or Apparatus 2 (paddle) may be more useful at higher rotation rates (e.g., the paddle at 100 rpm). Apparatus 3 (reciprocating cylinder) has been especially useful for bead-type modified-release dosage forms. Apparatus 4 (flow cell) may offer advantages for modified-release dosage forms that contain active ingredients that have limited solubility. Apparatus 7 (reciprocating disk) is applicable to nondisintegrating oral modified-release dosage forms, as well as to transdermal dosage forms. Apparatus 5 (paddle over disk) and Apparatus 6 (cylinder) also are useful for evaluating and testing transdermal dosage forms.

At least three timepoints are chosen to characterize the in vitro drug release profile of an extended-release dosage form for Pharmacopeial purposes. Additional sampling times may be required for drug approval purposes. An early time point, usually 1–2 h, is chosen to show that dose dumping is not probable. An intermediate time point is chosen to define the in vitro release profile of the dosage form, and a final time point is chosen to show essentially complete release of the drug.

Case A applies to an original modified-release oral dosage form for a drug already marketed in an immediate-release dosage form and for which extensive pharmacokinetic/pharmacodynamic data exist.

Case B applies to a non-oral, modified-release dosage form of an already marketed active drug entity for which extensive pharmacokinetic and pharmacodynamic data exist.

Case A studies (omitting the food effects studies) are appropriate for the evaluation of a modified-release dosage form designed for a non-oral route of administration if the pattern of biotransformation to active metabolites is identical for the two routes. If the biotransformation patterns are different, then clinical efficacy studies should be performed with the modified-release dosage form. In addition, special studies may be necessary to assess specific risk factors related to the dosage form (e.g., irritation and/or sensitization at the site of application of a transdermal drug delivery system).

Case C applies to a generic equivalent of an approved modified-release dosage form, which should be BE to the reference drug in its rate and extent of drug exposure (i.e., AUC, Cmax, Cmin, and degree of fluctuation) in crossover single-dose studies. For an oral modified-release dosage form, the food studies described under Case A also should be performed.

Case D applies to an FDA-approved product that has undergone SUPAC. Necessary in vitro dissolution tests and/or in vivo bioequivalence tests are described in the FDA guidance, SUPAC-MR: Modified Release Solid Oral Dosage Forms; Scale-Up and Postapproval Changes: Chemistry, Manufacturing, and Controls, In Vitro Dissolution Testing, and In Vivo Bioequivalence Documentation (www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM070640.pdf).

This level is the highest category of correlation. It represents a point-to-point relationship between in vitro dissolution and the in vivo input rate (absorption rate of the drug from the dosage form). For a Level A correlation, a product's in vitro dissolution curve is compared to its in vivo input curve, i.e., the curve produced by deconvolution of the plasma profile. Deconvolution can be accomplished using mass balance model-dependent methods, such as the Wagner–Nelson or Loo–Riegelman methods, or by model-independent, mathematical deconvolution. In an ideal correlation, the in vitro dissolution and in vivo absorption rate curves are superimposable or can be made superimposed by the use of a constant offset value of the time scale. The equations describing each curve are the same. This procedure often is found with modified-release dosage systems that demonstrate an in vitro release rate that is essentially independent of the dissolution media and stirring speeds used in a dissolution apparatus. Superimposition is not an absolute requirement for a Level A correlation. If the dissolution and absorption curves are different and a mathematical relationship can be developed to relate the two, the plasma level profile still is predictable from the in vitro dissolution data. This relationship must be true not only at that single input rate but also over the entire quality control dissolution range for the product. Furthermore, when the dissolution rate depends on mixing speed, the two curves can be made to superimpose by either increasing or decreasing the in vitro mixing speed or some other alteration of the dissolution method.

The advantages of a Level A correlation are as follows.

This correlation uses the principles of statistical moment analysis. The mean in vitro dissolution time is compared to either the mean residence time or the mean in vivo dissolution time. As with a Level A correlation, Level B uses all of the in vitro and in vivo data but is not considered a point-to-point correlation. It does not correlate the actual in vivo plasma profiles but rather a parameter that results from statistical moment analysis of a plasma profile component such as mean residence time. Because a number of different plasma profiles can produce similar mean residence time values, one cannot rely on a Level B correlation alone to predict a plasma profile from in vitro dissolution data. In addition, in vitro data from such a correlation cannot be used to justify values at the extremes of quality control standards.

This category relates one dissolution time point (t50%, t90%, etc.) to one pharmacokinetic parameter such as AUC, Cmax, or Tmax. It represents a single-point correlation and does not reflect the complete shape of the plasma profile, which best defines the performance of modified-release products. Because this type of correlation is not predictive of actual in vivo product performance, generally it is useful only as a guide in formulation development or as a production quality control procedure. Because of its obvious limitations, a Level C correlation has limited usefulness in predicting in vivo drug performance and is subject to the same caveats as a Level B correlation in its ability to support product and site changes as well as justification of the extreme values in quality control standards. The FDA Guidance “Extended-Release Solid Oral Dosage Forms—Development, Evaluation, and Application of In Vitro/In Vivo Correlations” (www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM070239.pdf) states that manufacturers can obtain biowaivers based on multiple Level C correlations. The guidance shows how manufacturers can achieve this correlation. The FDA also indicates that if such a correlation is achievable, it is likely that the development of a Level A correlation is also feasible.

This chapter does not define the only procedures for developing an IVIVC, and any well-designed and scientifically valid approach is acceptable. To assist the pharmaceutical scientist, one possible procedure for developing a Level A correlation is described below:

Reasonable upper and lower dissolution values are selected for each time point established from the biobatch. Historically, dissolution specifications have been selected by using the average dissolution of the development batches, with a range of ±2.5–3 standard deviations. It is now expected that the average dissolution values be approximately the same as those of the biobatch. The dissolution curves defined by the upper and lower extremes are convoluted to project the anticipated plasma level curves that would result from administration of these formulations to the same patients to whom the biobatch was administered. If the resulting plasma level data fall within the 95% confidence intervals obtained in the definitive BA-BE study, these ranges can be considered acceptable. An alternative acceptance approach that can be used after the therapeutic window for a drug has been defined, is to establish whether the upper and lower limits of the convolution results fall within the therapeutic window, even if they fall outside the confidence interval. If they fall outside the window, a more limited range must be established. This procedure should be continued until the predicted values meet the desired ranges.

An acceptable set of plasma-level data is established both for a batch of material demonstrating a more rapid release and for one demonstrating a slower release than that of the biobatch. These can be selected by using the extremes of the 95% confidence intervals or ±1 standard deviation of the mean plasma level. These curves are then deconvoluted, and the resulting input rate curve is used to establish the upper and lower dissolution specifications at each time point. In the case of Level B and C correlations, batches of product must be made at the proposed upper and lower limits of the dissolution range, and it must be demonstrated that these batches are acceptable by a BA-BE study.

Because the mechanisms for drug release from modified-release dosage forms are more complex and variable than those associated with immediate-release dosage forms, one would anticipate that an IVIVC would be easier to develop with the latter formulations. Unfortunately, most of the correlation efforts to date with immediate-release dosage forms have been based on the correlation Level C approach, although there also have been efforts employing statistical moment theory (Level B). Although it is conceivable that the same Level A correlation approach can be used with immediate-release dosage forms, until data have been gathered to support this concept, Level B and Level C are the best approaches that can be recommended with these dosage forms.