THE ROLE OF IN VITRO-IN VIVO CORRELATIONS (IVIVC) TO REGULATORY AGENCIES

Henry J. Malinowski

Director Division of Pharmaceutical Evaluation I Office of Clinical Pharmacology and Biopharmaceutics Food and Drug Administration 1451 Rockville, Maryland 20852

1. INTRODUCTION

24

Dissolution testing remains a potentially powerful and nearly always useful method for obtaining data related to quality and, potentially, clinical performance of dosage forms, especially solid oral dosage forms. But, not surprisingly, dissolution is not always a surrogate for bioequivalence, which necessitates human testing for determination of bio­equivalence in many instances. The key to confidence in dissolution testing is the strength of the relationship between dissolution and bioequivalence, in other words, the ability of dissolution testing to predict in vivo performance. Therefore, the availability of an IVIVC, as well as the type of dissolution testing conducted, are important considerations.

This discussion will put forward several examples of the use of IVIV relationships, including, but not limited to IVIVC, in various regulatory agencies around the world. Spe­cific IVIVCs and the application of IVIVCs remains a relatively unusual circumstance. Such correlations seem most likely for some ER drug products which leaves a large body of dosage forms, both Immediate release (IR) and extended release (ER), for which IVIVC are not available, for which alterative approaches have been utilized. These approaches in­clude concepts of side batches, the Biopharmaceutics Classification System and multi-me­dia dissolution testing. Also available are the "FIP Guidelines for Dissolution Testing of Solid Oral Products" Final Draft (1) and the FDA draft Guidance "Extended Release Solid Oral Dosage Forms - Development, Evaluation and Application of In Vitro/In Vivo Corre­lations" (2).

While IVIVC for ER products remains the optimal goal, other approaches currently can provide means for increased confidence in dissolution testing as a surrogate for bio­equivalence testing.

In Vitro--in Vivo Co"eiations, edited by Young et al. Plenum Press, New York, 1997 261

262 H. J. Malinowski

2. EUROPE

One recent, primarily European, effort which provides insight into current thinking in the area of IVIVCs is the final draft "FIP Guidelines for Dissolution Testing of Solid Oral Products." This guideline is expected to be finalized in late 1996. An excellent sum­mary of this Guideline was presented (3) by Martin Siewert. In the Guideline, a new term "in vitro-in vivo comparison" is used to identify a wider understanding than IVIVC or as­sociation. An in vitro-in vivo comparison is the process of comparing dissolution data to bioavailability data to determine what relationship exists between these two parameters. The stated purpose of these comparison studies is the "scientific verification of the in vitro test system and the respective specification limits for a given drug formulation." This comparison is applicable to both IR and ER dosage forms. It is useful to understand that, in the terminology of the Guideline, the comparison study may define a significant in vi­tro-in vivo association (lVIVC) but that useful information may still be obtained even when a correlation in the strict sense is not found.

For IR products, the comparison study suggested in the Guideline (Figure 1) For ER products, the Guideline endorses the categorization of correlation methods described in the USP, namely Levels A, B, and C. The type of correlation being attempted will deter­mine how many batches should be included in the correlation study. It is suggested that a single batch may be sufficient for an acceptable IVIVC only for a Level A correlation for a drug product with dissolution completely independent of environmental conditions. It is suggested that a Level A correlation can be used as a surrogate for bioequivalence testing for changes in manufacturing site, minor formulation changes, scale-up considerations as well as for setting dissolution specifications.

For ER products, two alternative methods are suggested for verification of dissolu­tion specifications, when a strict IVIVC cannot be developed. These are, situations where rank order correlation (Figure 2) is found and the concept of bioequivalence of side batches (Figure 3).

Dissolution [%] < in vivo comparison>

100

25~~~r-----1~----r---~

30 60 Time [m]

90 120 0 30 60 Time [m]

90 120

Figure 1. Specification type and verification study design concepts for immediate release products with Q speci­fied for grater than 15 minutes (used with permission) consists of a two-way crossover study between an oral solu­tion and a formulation which dissolves close to the dissolution specification limit.

The Role of in Vitro--in Vivo Correlations (IVIVC) to Regulatory Agencies 263

Bioavailability

lC~e~ ~ _.,·0. )

~-.. ----.·_.,.C ~ m Dissolution

) Specification limits

In Vitro Dissolution

Figure 2. Application of a rank order correlation for verification of in-vitro dissolution specifications ( used with permission).

In both of these situations, a separate bioequivalence study is suggested, to demon­strate bioequivalence of formulations with dissolution profiles near the upper and lower dissolution specifications.

3. JAPAN

For a Japanese view in this regard, I will refer a presentation (4) by Dr. Nobuo Aoy­agi from the Ministry of Health and Welfare (MHW) in Japan. This presentation focuses on the role of dissolution tests for bioequivalence assessment and consistently emphasizes the use of several dissolution conditions particularly with regard to pH. This relates to concerns associated with individuals exhibiting achlorhydria.

(Figure 4) shows simulated expected performance for 2 products, one of which ex­hibits pH dependent dissolution, while the other's dissolution characteristics are pH inde-

Dissolution [%)

100~~~~~~~l[==~~2:::~~E::=~~~~==~

2 4 6 Time [h)

8

24

10 12

Figure 3. In-vivo verification of in-vitro test systems and specification based on the side-batch approach (exam­ple)( used with permission).

264

Tablet A

pH-dependent dissolution

Tablet B

pH-independent dissolution

Subject N1 ormal acidity

GET=lh

Subject N3 Normal acidity GET=3h

H. J. Malinowski

Subject L Hypoacidity

lC~L 6 0

IOO~ l~ ~ ~ - - pH 1 ~ - pH7

o 2 • T,me (h)

6 0

6 0

Figure 4. In vitrolin vivo dissolution profiles in three subjects with different gastric acidity and gastric emptying time for tablets A and B showing pH-dependent and independent dissolution (used with permission).

pendent. Shown are both dissolution and in vivo absorption data which indicate that the product with pH independent dissolution will more easily matches the expected bioavail­ability results in both patients with normal acidity or hypoacidity. For the pH dependent dissolution product, no one set of dissolution conditions will describe expected bioavail­ability for all patients. An important point to be made is that a product which shows pH independent dissolution in vitro is probably also pH independent, as far as dissolution, in vivo. This type of product is likely to be less susceptible to gastric acidity differences among patients.

Another aspect of the process establishing a relationship between dissolution and bioavailability is shown in Figure 5.

This shows bioavailability data for 2 subjects for 2 ER products. In one subject, the bioavailability profiles are very similar, in the other subject, the profiles are quite differ­ent. Somewhat of a relationship is shown for subject K using basket 150 rpm conditions, but, for Subject N, only very unusual conditions, paddle 0 rpm (1 minute stirring at each sampling time) resulted in a in vitro in vivo relationship (IVIVR). This example illustrates that in vivo data can be quite variable, as seen in these 2 subjects. It is not possible, or rea­sonable, to try to establish an IVIVR for each individual subject. Therefore, average data are normally used for establishing such relationships.

The importance of pH conditions, both in vitro and in vivo, is shown in Figure 6. These 2 diazepam products both dissolve very rapidly in pH1.2 dissolution medium. And in subjects with normal gastric acidity, both products have similar bioavailability. How­ever, in subjects with hypoacidity, bioavailability differences are apparent and this is re­flected in dissolution testing using pH 4.6 medium. This further illustrates the importance attached to studying several pH ranges for dissolution testing, related to concerns about gastric hypoacidity, by regulatory authorities in Japan. A detailed description of recom­mended dissolution testing may be found in the recent draft "Guideline for Bioequivalence Testing" (5) for generic drugs from the MHW.

The Role of in Vitro-in Vivo Correlations (IVIVC) to Regulatory Agencies

Subject N Subject K 100 In vitro : Paddle, 0 rpm

2 4 6 o 2 4 Time (h) Time (h)

6

CR·B vivo CR·B vitro

CR·A vitro

CR·A vivo

265

Figure 5. In vivo release of acetaminophen from CR-A and B in typical two subjects and in vitro release by JP paddle method a 0 rpm (I min stirring at each sampling time) and rotating basket (RB) method at 150 rpm (used with permission).

Figure 7 illustrates the distribution of GI variables throughout the population. This is shown in relation to the degree of discrimination of tests for bioequivalence assurance, ranging from discriminatory (perhaps over discriminatory) conditions to nondiscrimina­tory conditions. It suggests that tests to assure bioequivalence generally are quite discrimi­natory, focused not on the average individual, but including nearly everyone in the population. While such typical conservative test conditions do provide a safety margin in detecting differences in measured parameters, they can also lead to situations where dif-

80

60

-g > (5 '" 40 '" U ~

20

o

pH 1.2 pH 4.6

--A -<>-C

o 10 0 10 20 JO 40 50

Time (min)

JOO

~ 200 E 0, .S­o c: o () 100

Normal acidity Hypoacidity

o O---'-_-L---'_...J o 2 4 680 2 4 6 8

Time (h) Time (h)

Figure 6. In vitro dissolution of diazepam from tablets A and C by a beaker method and serum concentration in two groups of subjects with normal and hypo·acidity of gastric fluid.

266 H. J. Malinowski

Distribution of GI variables

~ Overestimation range

-----Test for BE assurance

"U hb A j .... Tat>B 'it - .

T",,"

Physiological range

Discriminatory condition - ------------

Underestimation range

No ndi scri m ina tory condition

Figure 7. Physiological range of critical GI variables and in vitro testing condition ofbioequivalence (BE) assur­ance

ferences are detected in, for example, dissolution testing, which do not relate to differ­ences in bioavailability.

4. NORTH AMERICA

In the United States, a draft (2) Guidance for Development, Evaluation and Applica­tion of in Vitro/in Vivo Correlations was released on July 10, 1996. This guidance pro­vides recommendations to pharmaceutical scientists related to various aspects of IVIVC for oral extended-release (ER) drug products particularly as utilized in the NDA/ANDA review process. It presents a comprehensive perspective on methods of developing IVIVC, appropriate means of evaluating the predictability of IVIVC, and relevant applications for IVIVC in the areas of changes (e.g., formulation, equipment, process, and manufacturing site) and setting dissolution specifications.

In vitro dissolution testing is important for: (1) providing necessary process control and quality assurance, (2) determining stability of the relevant release characteristics of the product, and (3) facilitating certain regulatory determinations and judgments concern­ing for example, minor formulation changes or change in site of manufacture. In addition, in certain cases, especially for ER formulations, the dissolution test can serve not only as a quality control for the manufacturing process but also as an indicator of how well the for­mulation will perform in vivo. Thus, the main objective of developing and evaluating IVIVC is to empower the dissolution test to serve as a surrogate marker for human bio­equivalence studies. One additional purpose of establishing an IVIVC is to minimize the number of human studies needed for approving and maintaining a drug product on the market. This approach will not only reduce the cost and time of drug development by re­ducing the number of studies required to demonstrate adequate bioavailability, but will also facilitate the initial approval as well as scale-up and post-approval changes. However, for certain applications the adequacy of the in vitro method to act as a surrogate for in vivo

The Role of in Vitro-in Vivo Correlations (IVIVC) to Regulatory Agencies 267

testing must be demonstrated through an IVIVe for which predictability has been demon­strated.

4.1. Biowaivers

Regarding biowaivers, five categories of waivers are described in the FDA Guid­ance. These range from situations which are insignificant as far as expected effect on product performance, such as a change in manufacturing equipment where the new equip­ment has the same design and operating principles, to very significant changes for which biotesting would be required even if an IVIve with good predictability has been devel­oped. An example of a very significant change situation is the approval of another spon­sor's ER product, for which an IVIVe has not been specifically developed, even with the same release controlling mechanism, where the reference product does have an IVIVC es­tablished.

Between these 2 categories are three categories for I) non-narrow therapeutic index drugs, 2) narrow therapeutic index drugs and 3) ER drug products which have dissolution characteristics which are independent of dissolution test condition. Specific recommenda­tions related to IVIve for changes in site of manufacture, release-controlling and non-re­lease-controlling excipients, manufacturing equipment and process, as well as approval of certain new strengths, are categorized and described in the draft Guidance, for each of these categories.

The criteria for granting biowaivers in circumstances where a IVIVe has been estab­lished are that the difference in predicted means of e max and Aue is no more than 20% from that of the reference product and, where applicable, the new formulation meets the application/compendial dissolution specifications.

4.2. Setting Dissolution Specifications

One additional important use of IVIVC, as described in the draft Guidance, relates to using the IVIve in the process of setting dissolution specifications. Also described is the situation of setting dissolution specifications where there is no IVIVC.

4.2.1. No IV1VC. In general, USP acceptance criteria for dissolution, regarding set­ting specifications, are utilized unless alternate acceptance criteria are justified. One less than optimal approach is to have the specifications established such that all lots pass at Stage I of testing. In other words, each individual dosage form among the dissolution data being utilized to determine appropriate specifications, passes the proposed specifications. Therefore, it is recommended that specifications should be established based on average dissolution data (Stage 2). Specification ranges of 20% or less are recommended. If justi­fied, deviations from this criteria can be accepted, up to ranges of approximately 25%.

Specification ranges greater than 20% are generally acceptable only when supported by evidence that lots with mean dissolution profiles that are allowed by the upper and lower limit of the specifications are bioequivalent.

4.2.2. When an 1V1VC Has Been Established. An important application of an IVIVe is in relation to setting appropriate dissolution specifications for the product. Ideally, the specifications should be established such that all lots that have dissolution profiles within the upper and lower limits of the specifications, are bioequivalent. Minimally, these lots should be bioequivalent to the lots used in the clinical trials or an appropriate reference

268 H. J. Malinowski

standard. Specifications should be set on mean data using at least 12 individual dosage units per data set. Calculate the plasma concentration time profile for the upper and lower proposed dissolution specification profiles and using convolution techniques (or other ap­propriate modeling techniques) determine whether the lots with the fastest and slower re­lease rates that are allowed by the dissolution specifications result in a maximal difference of 20% in the predicted Cmax and AUC.

5. CONCLUSION

IVIVCs are not extensively used at this time by regulatory agencies around the world. However, there is general agreement that IVIVCs are very useful and desirable. And, IVIVRs, that is, a dissolution test which is meaningful in the sense that some rela­tionship between the dissolution test results and expected bioavailability changes has been established, are universally accepted.

Recent draft Guidelines from Japan, the Federation Intemationale Pharmaceutique (FIP), and the U.S. FDA can provide much useful information regarding current thinking in the major drug regulatory jurisdictions around the world regarding the role of in vitro/in vivo correlations as well as other in vitro/in vivo relationships.

REFERENCES

1. Guidelines for Dissolution Testing of Solid Oral Products Final Draft 1995, Phannacopoeial Forum, 21 (5), 1371-1382,1995.

2. Guidance for Industry - Extended-Release Solid Oral Dosage Fonns: Development, Evaluation and Appli­cation of in Vitro/in Vivo Correlations, Final Draft July I, 1996. Food and Drug Administration.

3. New FIP Guideline for Dissolution Testing of Solid Oral Products, Dissolution Technologies 3 (3), 3--6, 1996.

4. Use of Dissolution Tests for Bioequivalence Assessment in Japan, Nobuo Aoyagi, Abstract, FIP Bio Inter­national 1996, 132-135.

5. Guideline for Bioequivalence Studies of Generic Drugs, Draft July I, 1996, Ministry of Health and Wel­fare, National Institute of Health Sciences, Japan.