`Printed in Denmark . All rights reserved
`
`Copyright C
`ISSN 1742-7835
`
`Short Communication
`
`Dissolution – Bioequivalence Non-Correlations
`
`Constantin Mircioiu1,2, Ion Mircioiu2, Victor Voicu1 and Dalia Miron1,2
`1University of Medicine and Pharmacy ‘‘Carol Davila’’, Bucharest, and 2Biopharmacy & Pharmacol Research S.A.,
`Bucharest, Romania
`
`The World Health Organization (WHO) rules (1996) rec-
`ommend in chapter 3 ‘‘Technical data for regulatory assess-
`ment’’ that information for marketing authorization should
`contain among others in-documentation, equivalence data
`(comparative bioavailability, pharmacodynamic or clinical
`studies), and comparative in vitro dissolution tests. Euro-
`pean rules concerning evaluation of bioavailability and bi-
`oequivalence (CPMP/EWP/QWP/1401/98) specify in chap-
`ter 5.2:
`‘‘Dissolution studies are always necessary and
`consequently required. In vitro dissolution testing forms a
`part of the assessment of a bioequivalence waiver’’. The
`consequence is that over the last years all over Europe dis-
`solution studies were connected to bioequivalence studies
`and the natural tendency was to correlate the results ob-
`tained in the pair studies in order to obtain models allowing
`dissolution results as predictor for in vivo results.
`Let us consider only as a mnemotechnique, the living
`body as a ‘‘mathematical operator’’ which transforms par-
`ameters characterizing dissolution curves in pharmaco-
`kinetic parameters associated to plasma levels curves.
`
`y (dissolution) Ω pharmacokinetics
`
`If we consider the two spaces as ordered metric spaces, a
`natural question is that the operator preserves the distances
`and order. For example, if we compare a tested drug T with
`a reference one R, we would be interested to know if
`
`din vitro (R, T) Ω din vivo (R, T)
`
`and RⱕT⇒y(R)ⱕy(T)
`
`Less formally speaking we are interested to know if in vitro
`similarity implies bioequivalence and if a better dissolution
`implies a better bioavailablity. It is clear that the response
`depends on the active substance and physiopatological par-
`ameters but also on the particular metrics choiced for char-
`acterizing in vitro and in vivo curves as well as distances
`between its. The response is not so simple, first of all since
`
`Author for correspondence: Constantin Mircioiu, Faculty of Phar-
`macy, Traian Vuia 6, Bucharest, Romania (fax π4021 6101550,
`π4021 3101410, e-mail cmirc/gg.unibuc.ro).
`
`due to the complexity of the phenomena studied, both met-
`rics and order in vitro and in vivo are not well defined. The
`great number of dissolution and bioequivalence metrics
`(Enachescu et al. 2003) show that the problem is yet to be
`solved.
`In vitro dissolution tests. Dissolution tests were performed
`using the method indicated by USP or according to the
`specifications provided by the producers. As metrics of dis-
`solution were considered the factors f1 and f2.
`Clinical trials. Each study was performed on healthy vol-
`unteers. Experiments were of the standard type: cross-over,
`with two periods and two sequences.
`Analytical methods. Plasma levels of the drugs were evalu-
`ated using validated liquid chromatographic methods, with
`UV or mass spectometry detection.
`In judging the results it should be kept in mind that that
`in so-called in vitro/in vivo correlations, we practically jump
`over one step – in vivo dissolution, which is by far more
`variable and more complex than the in vitro dissolution.
`Since in vitro dissolution conditions are often far from the
`in vivo conditions, we have non-correlation between the two
`dissolutions.
`Similar dissolution and non-bioequivalence. This is the
`case for many acidic or basic drugs. Most representative
`is mefenamic acid (fig. 1). If we calculate factors f1 or f2,
`dissolution curves are similar. The curves are practically
`much more similar that these factors indicate, negative and
`positive areas between the curves being approximately
`equal. Consequently, if we chose as norm of dissolution
`curves area under experimental data (AUC), the distance
`between the two curves d(ref, test) Ω | AUCrefªAUCtest | is
`approximately zero. In spite of this high similarity in dissol-
`ution, plasma leves are quite different when it concerns cmax
`and AUC.
`Non-correlation issues from the fact that in vitro release
`medium had a pHΩ8, which is far from physiological con-
`ditions. Since mefenamic has a very low solubility in acidic
`and neutral media following its lipophilic, week acid charac-
`ter, the alkaline medium was chosen by the producer in or-
`der to obtain a ‘‘good dissolution’’ in vitro, without con-
`sidering to a correlation with in vivo release conditions.
`
`Page 1
`
`SHIRE EX. 2031
`KVK v. SHIRE
`IPR2018-00293
`
`
`
`SHORT COMMUNICATION, C Pharmacology & Toxicology 2004, 95, 262–264.
`
`Fig. 1.
`
`Fig. 2.
`
`Page 2
`
`
`
`SHORT COMMUNICATION, C Pharmacology & Toxicology 2004, 95, 262–264.
`
`Non-similar dissolution but bioequivalence. This is quite a
`frequent case for many molecules of different structures.
`One explanation could arise from the fact that in vitro non-
`similarity is connected with a somewhat 10% difference (for
`all pairs of points of dissolution curves there is 10% differ-
`ence, i.e. f2Ω50) and in vivo non-bioequivalence is about a
`20% difference. This is a very roughly characterization for
`the acceptance limits of dissolution and bioequivalence, but
`the idea that dissolution metrics are more refined that bioe-
`quivalence metrics deserves much more attention.
`In some cases, the mechanism may be more specific, for
`example in the case of methotrexate (fig. 2).
`The reference and the product tested attained approxi-
`mately superposable mean plasma levels curves, though in
`vitro dissolution curves were dissimilar whatever the dissol-
`ution metrics used to measure their distance. Since metho-
`trexate has a great molecular weight, a free diffusion mech-
`anism of absorption is less probable. An active transport
`was not described. An absorption via embedding in micelles
`of physiological surface active agents (mainly bile acids) re-
`mains a more reliable hypothesis. If formation and transfer
`of micelles across intestinal barrier is the rate-limiting step
`of the entire process (in vivo release and absorption),
`quenching of differences in dissolution could appear.
`Simvastatin was also in this category of non-correlation,
`
`but in this case there is clearly a problem of metrics of dis-
`solution. As discussed above, since the areas under dissol-
`ution curves were approximately equal, it is more reason-
`able to think that the products have similar dissolution
`though all usual metrics argue against this idea.
`Conclusion. Though helpful, the use of a comparative dis-
`solution in prediction of in vivo bioequivalence offers some
`problems and the risk of false predictions should be kept in
`mind. The use of acidic dissolution medium for basic drugs
`or basic medium for acidic drugs lead to more optimistic
`estimations than the actual situation. In upcoming rules for
`prediction of in vivo behaviour from dissolution results we
`need more knowledge about adequate tests for use and
`comparison of tests made under different conditions.
`
`References
`
`WHO Expert Committee on Specifications of Pharmaceutical Prep-
`arations. Thirty-fifth report. Geneva, World Health Organiza-
`tion. WHO Technical Report Series, 1996, No. 863, Annex 9,
`chapter 3.
`Enachescu D., C. Enachescu, C. Mircioiu & I. Mircioiu: On the
`classification of sets of experimental points and curves in bi-
`opharmacy. Dissolution metrics. Craiova Med. J. 2003, 5, supl 3,
`493–496.
`
`Page 3
`
`