http://informahealthcare.com/phd ISSN: 1083-7450 (print), 1097-9867 (electronic) Pharm Dev Technol, Early Online: 1–4 ! 2015 Informa Healthcare USA, Inc. DOI: 10.3109/10837450.2015.1035726
SHORT REPORT
About the equivalence between different batches of a glycopeptide drug Antonio Boix-Montan˜es1 and Alfredo Garcia-Arieta2
Pharmaceutical Development and Technology Downloaded from informahealthcare.com by Kainan University on 04/28/15. For personal use only.
1
Departamento de Farmacia y Tecnologia Farmace´utica, Facultad de Farmacia/Universidad de Barcelona, Barcelona, Spain and 2Divisio´n de Farmacologı´a y Evaluacio´n Clı´nica, Departamento de Medicamentos de Uso Humano, Agencia Espan˜ola de Medicamentos y Productos Sanitarios, Madrid, Spain Abstract
Keywords
Context: Teicoplanin is a glycopeptide antibiotic consisting of a combination of different active components. Clinical equivalence between different batches of this drug is not guaranteed by the present pharmacopeial specification of chemical composition based on an HPLC chromatogram. Objective: To test a modification of this specification and to evaluate independent compositions recently published in the literature. Materials and methods: The expectable area under the plasma curve of each batch has been estimated based on its chemical composition as described in a former paper. Batchcomparisons are based on ratios between the area of the test batch and the area of a reference. Results and discussion: The modification of this specification recently proposed by the European Medicines Agency (EMA) has been tested confirming its goodness. A new acceptance range of AUC variation, rounding 10% to +15%, has been obtained. It is narrower than the current interval of the pharmacopeial specification. Concerning the generic batches that have been studied, the majority of differences with the reference is lower than ±10%. Variations in the compositions of the reference product have been observed to influence the results and a control criteria are proposed. Conclusion: The variability of the pharmacokinetic performance of teicoplanin can be better controlled with this new proposal of composition specification given by EMA.
Analytical specification, pharmacopeia, relative bioavailability, teicoplanin
Introduction The majority of glycopeptide antibiotics, from fermentative origin, consists of a combination of closely-related molecules and its resulting activity depends on the individual contribution of each active component. Concerning Teicoplanin, a Vancomycinlike antibiotic, the active subcomponents are identified as the ‘‘A2’’ group. All pharmacopeias describe the same five A2 subcomponents that are numbered based on a rank of retention time values. Concerning the chemical differences between each subcomponent, the molecule core is the same for all of them, but each one contains a slightly different aliphatic radical chain giving slight differences in polarity to the whole molecule1. Consequently, their specific pharmacokinetic characteristics are directly related with their physicochemical properties2. Although injectable generic drug applications to regulatory agencies are generally waived from pharmacokinetic bioequivalence studies3, in the case of teicoplanin there is a dearth of evidence with respect to the equivalent systemic exposure between different pharmacopeia-compliant batches4,5.
Address for correspondence: Dr. Antonio Boix-Montan˜es, Departamento de Farmacia y Tecnologia Farmace´utica, Facultad de Farmacia/ Universidad de Barcelona, Av. Joan XXIII, 08028 Barcelona, Spain. Tel: +34 934024560. Fax: +34 944024563. E-mail: [email protected]
History Received 9 December 2014 Revised 30 January 2015 Accepted 6 March 2015 Published online 17 April 2015
Assay specifications of this antibiotic in the European and Japanese pharmacopeias6,7 comprise the microbiological potency of the whole substance and also the chromatographic composition of A2 subcomponents. In this sense, the European Medicines Agency (EMA) has recently recommended8 to constrain the current specification of the chromatographic composition. This article examines how to find a scientific rationale to this modification using a practical case. Brink et al.9 pointed out that significant differences in A2i compositions (having a higher percentage of hydrophilic subcomponents) may lead to a lower sum of the areas under the teicoplanin plasma concentration versus time curve (AUC) of the A2i subcomponents. In addition, Rossi et al.10 have performed a principal component analysis (PCA) of a collection of A2i compositions of reference and generic batches with an ad hoc statistical model stating that this uncertainty between batches could only be solved by quantifying the effect on tissue drug concentrations. Certainly, it would be the best demonstration if an accurate knowledge of those levels would be achievable. In practice, a more suitable approach is needed to compare differences between marketed batches. A comparison based on known pharmacokinetic differences of each A2 subcomponent is a more reliable approach although this information is scarcely available in the literature. As an estimator of pharmacokinetic exposure, very few authors have published the AUC of the individual A2 subcomponents in relation with the
Pharmaceutical Development and Technology Downloaded from informahealthcare.com by Kainan University on 04/28/15. For personal use only.
2
A. Boix-Montan˜es & A. Garcia-Arieta
Pharm Dev Technol, Early Online: 1–4
respective HPLC composition of this injectable formulation1,11. Assuming this scarcity of useful data, we have related in a previous paper12 the experimental A2i compositions of a set of teicoplanin batches (normalized by the corresponding amount originally infused) with their respectives AUCs taken from the literature. In clinical practice, the AUC of teicoplanin is estimated as the linear combination of the contribution of each A2 subcomponent. In our case, the expected whole AUC of any teicoplanin batch has been estimated as the linear combination of these calculated relationships weighted by the respective composition of each batch of substance. As a result, some generic batches were compared based on this prediction of systemic exposure. The aim of this work is to try out this calculation approach to assess the pharmacokinetic significance of the specification proposed by EMA and also the relevance of the inter-batch differences of this drug using recently published sets of real data.
Methods A set of A2 compositions of 14 generic batches of teicoplanin that were published in the literature10 has been used to estimate their expected AUCi values using the individual response factors calculated as detailed in Ref. 12. A single AUC value has been calculated for each batch as the arithmetic sum of the contributions of each A2 component1. Individual AUCs have been calculated from the A2 composition using an AUC response factor of each subcomponent weighted by its respective percentage in the teicoplanin composition in each batch (AUCTEST). Two different sets of teicoplanin reference compositions have been comparatively used in this study to calculate a representative AUC value for the reference (AUCREF): (1) A pool of six batches from four different laboratories (24 A2 compositions)12. (2) A set of 11 batches of the reference product that were analyzed simultaneously with the 14 generic batches under study10. The corresponding mean values of each series were named as Ref1 and Ref2, respectively. Finally, the corresponding ratios AUCTEST/AUCREF for each generic batch have been calculated. Two series of results have been obtained using either Ref1 or Ref2 and their differences have been discussed. The obtained AUC ratios express the deviation of the AUC of the test batch (i.e. a generic batch) from the AUC given by a reference composition. In a second investigation, the estimations of AUCs have been used to estimate the inter-batch variability in A2 composition using the largest set of values available from a single manufacturer: Ref2. For this purpose, individual AUC values of Ref2 have been used as test values (numerator) and their mean has been used as the reference AUC (denominator).
Finally, the new specification proposed by EMA has been tested by calculating its corresponding hypothetical extreme compositions in A2 subcomponents. The theoretically lowest AUC would result from the maximum possible content of the most polar components (A2-1, A2-2, etc.). Meanwhile, the highest AUC would result from the maximum acceptable content of the most lipophilic ones (A2-5, A2-4, etc.). These ‘‘side-batches’’ have been taken to calculate the corresponding AUCTEST values of the extremes that are implicitly accepted by this specification. A summary of those values is shown in Table 1.
Results Mean AUC values (standard deviations within parentheses) for each reference data set are 325.71 (2.57) mg/mL h for Ref1 and 313.89 (8.50) mg/mL h for Ref2. Table 2 summarizes the estimated AUC values of the 14 test batches and resulting ratios using both Ref1 and Ref2 mean values. Using Ref1, the AUCs of 8 of 14 generic batches would differ more than ±5% of the reference and 3 of 14 batches would differ more than ±10%. Considering Ref2, 6 of 14 fall outside ±5% and 3 of 14 fall outside a ±10% level. In both cases, batch UKB10 turned out to be the worst case. It is shown that the equivalence of a generic batch with a reference composition depends not only on the test value but also on the specific value taken as the reference. Concerning the variability of the set of Ref2 batches, 3 of 11 batches fall outside a ±5% range and none of them outside the ±10% of their mean value. Resulting AUC ratios of each side-batch with the corresponding mean AUCREF range between 89.03% and 116.31% (from 10.97% to +16.31%) with Ref1 and between 92.39% and 120.71% (from 7.61% to +20.71%) with Ref2.
Table 2. Individual AUC values (mg/mL h) and their corresponding ratios for each generic batch considering references Ref1 and Ref2. Batch AB951 CO587 CO608 CR111 E275A FE004 GZ802 GZ805 K6802 LA103 LA104 LA105 UE821 UKB10
Estimated AUC
Ref1
Ref2
321.72 307.46 311.28 291.52 304.20 299.27 344.93 328.33 349.50 333.98 322.10 324.75 276.70 265.64
98.77 94.39 95.56 89.49 93.39 91.87 105.89 100.79 107.29 102.53 98.88 99.70 84.94 81.55
102.50 97.95 99.17 92.87 96.91 95.34 109.89 104.60 111.35 106.40 102.62 103.46 88.15 84.63
Bold numbers indicate deviations higher than ±5%.
Table 1. Specifications of A2 composition (%) of the European pharmacopeia (Reference 7) and of the EMA proposal (Reference 8) and their corresponding theoretical side-batches. Reference European pharmacopeia v. 7.0 Eur low side-batch Eur high side-batch EMA proposal EMA low side-batch EMA high side-batch
A2–1
A2–2
A2–3
A2–4
A2–5
Sum
0–20.0 20.0 0 10.0–19.0 19.0 10.0
35.0–55.0 55.0 40.0 37.0–50.0 46.0 40.0
0–20.0 5.0 20.0 5.0–11.0 5.0 11.0
0–20.0 0 20.0 7.0–15.0 7.0 15.0
0–20.0 0 20.0 7.0–17.0 7.0 17.0
80.0–100.0 80.0 100.0 84.0–93.0% 84.0 93.0
DOI: 10.3109/10837450.2015.1035726
Pharmaceutical Development and Technology Downloaded from informahealthcare.com by Kainan University on 04/28/15. For personal use only.
Discussion Normally distributed variations have been assumed for both sets of reference compositions. Considering that Ref1 includes the data from four different laboratories instead of one in the case of Ref2, inclusion of the interlaboratory variability results in a more representative estimation of the population of reference batches. We have assumed that Ref1 could be a better estimation of the true data. Using Ref1, the newly proposed EMA specification seems to imply the acceptance of maximum differences in AUC from around 10% to +15%. Meanwhile, the current European pharmacopeial specification leads to around a ±20% acceptation12, which is wider. Those results suggest that this new specification would improve the batch consistency. As expected, the modification of the specification of A2i composition suggested by EMA would control the therapeutic performance of teicoplanin better than the current pharmacopeial specification because the width of the acceptance range has been narrowed. Other proposals of A2 composition would be possible and could be assessed using this calculation approach based on the expected pharmacokinetic AUC. The noticeable variations of the reference value (Ref2) influence the result of the comparison. An approach to control this uncertainty could be to include a significant number (e.g. n 6) of A2 compositions of different reference batches in the same analytical run of the test batches and to accept the same variability that is accepted with the generics under testing, e.g. not more than ±10% departure from the mean value. Concerning the clinical relevance of the comparisons, it is known that principal components are defined as linear combinations of the original variables preserving the original variance13. The PCA of teicoplanin batches published by Rossi et al.10 searches for the maximum discrimination between reference and innovator batches to achieve a maximum separation among predefined classes. This actual combination means a feature selection instead of a feature reduction which should be the main goal of a PCA. In this sense, this mathematical approach is inappropriate to decide about clinical relevance of batch differences. In fact, drawing conclusions between batches should have to be based on biological considerations such as a pharmacokinetic performance or the expected antimicrobial efficacy of each batch. Nevertheless, a comparison based on MIC values has certain limitations as: MIC values found in the literature are discrete values (not continuous), data of the literature cannot be pooled (thorough information of analytical conditions is lacking), lower analytical selectivity than a chromatographic method, poorer predictability of this in vitro test compared with the in vivo pharmacokinetics, etc. Alternatively we have compared, with an agreed mean reference, all the individual values of this dataset of 14 generic batches on the basis of a pharmacokinetic descriptor of systemic exposure. As a result, the majority of AUC values of those batches differs less than ±5% of the reference composition, and few batches differ more than ±10%. Despite the scarcity of information in the literature about the relationship between A2 chemical compositions and its achievable plasma AUCs, it is possible to demonstrate a pharmacokinetically-based equivalence of a generic batch with a pool of well-standardized reference batches. It is also possible to predict the width of the acceptance range that corresponds to any A2 specification proposal. Differences among pharmacopeial compliant batches of other fermentative drugs have been also controversial regarding the microbiological or clinical comparability between different origins or manufacturers14,15, being attributed to differences in purification steps. If it was the case, further quality specifications should be proposed in addition to the chemical composition.
Equivalence between batches of a glycopeptide drug
3
It is worth noting that this article refers only to the specification of the chemical composition of teicoplanin and not to a conventional bioequivalence calculation. This comparison of the compositions of the active components based on bioavailability concepts has been shown to be useful to understand the modification of the current chemical specification values moving to more relevant ranges in terms of a better controlled pharmacokinetic exposure. In summary, this calculation approach has quantified the expected pharmacokinetic differences (described by AUC, the more suitable descriptor of bioavailability extent) due to inter-batch differences in chemical composition. It is remarkable to note that pharmacokinetic differences with the new specification would be lower than observed using current specifications. The acceptance range would be more restrictive and a more homogeneous systemic exposure of this antibiotic would be easily guaranteed.
Conclusions The prediction of the expected area under the plasma levels versus time curve of each active substance is a possible approach to control the risk of exposure variations of teicoplanin batches from different origins. This approach has been used to evaluate the pharmacokinetic performance of any specification of the chemical composition of A2 subcomponents and also to test the equivalence of different test batches with a pre-defined reference. This methodology can be applied to similar multicomponent antibiotics where a clinical rationale for the definition of its chemical composition is required.
Declaration of interest The authors report no conflicts of interest.
References 1. Bernareggi A, Danese A, Cometti A, et al. Pharmacokinetics of individual components of teicoplanin in man. J Pharmacokinet Biopharm 1990;18:525–543. 2. Rowland M. Clinical pharmacokinetics of teicoplanin. Clin Pharmacokinet 1990;18:184–209. 3. European Medicines Agency. Guideline on the investigation of bioequivalence. CPMP/EWP/QWP/1401/98 Rev. 1/Corr. 2010 Available from: http://www.emea.europa.eu/docs/en_GB/document_ library/Scientificguideline/2010/01/WC500070039.pdf [last accessed 13 Jun 2014]. 4. European Medicines Agency. Questions and answers on the referral for teicoplanin Hospira powder and solvent for injection containing 200 or 400 mg teicoplanin. EMA/CHMP/77992/2010. EMEA/H/A29/1084. 2009. Available from: http://www.ema.europa.eu/docs/ es_ES/document_library/Referrals_document/Teicoplanin_Hospira_ 29/WC500014116.pdf [last accessed 17 Feb 2013]. 5. Fujimura S, Fuse K, Takane H, et al. Antibacterial effects of brandname teicoplanin and generic products against clinical isolates of methicillin-resistant Staphylococcus aureus. J Infect Chemother 2011;17:30–33. 6. Japanese Ministry of Health, Labour and Welfare. Japanese Pharmacopoeia 16th ed. Teicoplanin monograph. 2012. Available from: http://jpdb.nihs.go.jp/jp16e/JP16.pdf [last accessed Nov 2012]. 7. European Council. European pharmacopoeia 7.0 corrected 6.6. Teicoplanin monograph (01/2009:2358). Available from: http:// www.edqm.eu/en/european-pharmacopoeia-publications-1401.html [last accessed Nov 2012]. 8. European Medicines Agency. Assessment report. Review under Article 5(3) of Regulation (EC) No 726/2004. Teicoplanin. Procedure no. EMEA/H/A-5(3)/1315. EMA; 21 March 2013. Available from: http://www.ema.europa.eu/docs/enGB/document library/Report/2013/04/WC500142229.pdf [last accessed 13 Jun 2014]. 9. Brink AJ, Richards GA, Colombo G, et al. Multicomponent antibiotic substances produced by fermentation, implications for
4
A. Boix-Montan˜es & A. Garcia-Arieta
Pharmaceutical Development and Technology Downloaded from informahealthcare.com by Kainan University on 04/28/15. For personal use only.
regulatory authorities, critically ill patients and Generics. Int J Antimicrobial Agents 2014;43:1–6. 10. Rossi A, Buttini F, Colombo P, et al. Complex product composition generates risks for generic substitution also with dosage forms for intravenous administration. Int J Pharm 2013;451:50–56. 11. Falcoz C, Ferry N, Pozet N, et al. Pharmacokinetics of teicoplanin in renal failure. Antimicrob Agents Chemother 1987;31:1255–1262. 12. Boix-Montan˜es A, Garcia-Arieta A. Composition specification of teicoplanin based on its estimated relative bioavailability. Drug Dev Ind Pharm 2015;41:218–223.
Pharm Dev Technol, Early Online: 1–4
13. Massart DL, Vandeginste BGM, Deming SN, et al. Chemometrics, a textbook. Data handling in science and technology. Amsterdam: Elsevier; 1988. 14. Vesga O, Agudelo M, Salazar BE, et al. Generic vancomycin products fail in vivo despite being pharmaceutical equivalents of the innovator. Antimicrob Agents Chemother 2010;54:3271–3279. 15. Agudelo M, Rodriguez CA, Pelaez CA, Vesga O. Even apparently insignificant chemical deviations among bioequivalent generic antibiotics can lead to therapeutic nonequivalence: the case of meropenem. Antimicrob Agents Chemother 2014;58:1005–1018.
SHORT REPORT
About the equivalence between different batches of a glycopeptide drug Antonio Boix-Montan˜es1 and Alfredo Garcia-Arieta2
Pharmaceutical Development and Technology Downloaded from informahealthcare.com by Kainan University on 04/28/15. For personal use only.
1
Departamento de Farmacia y Tecnologia Farmace´utica, Facultad de Farmacia/Universidad de Barcelona, Barcelona, Spain and 2Divisio´n de Farmacologı´a y Evaluacio´n Clı´nica, Departamento de Medicamentos de Uso Humano, Agencia Espan˜ola de Medicamentos y Productos Sanitarios, Madrid, Spain Abstract
Keywords
Context: Teicoplanin is a glycopeptide antibiotic consisting of a combination of different active components. Clinical equivalence between different batches of this drug is not guaranteed by the present pharmacopeial specification of chemical composition based on an HPLC chromatogram. Objective: To test a modification of this specification and to evaluate independent compositions recently published in the literature. Materials and methods: The expectable area under the plasma curve of each batch has been estimated based on its chemical composition as described in a former paper. Batchcomparisons are based on ratios between the area of the test batch and the area of a reference. Results and discussion: The modification of this specification recently proposed by the European Medicines Agency (EMA) has been tested confirming its goodness. A new acceptance range of AUC variation, rounding 10% to +15%, has been obtained. It is narrower than the current interval of the pharmacopeial specification. Concerning the generic batches that have been studied, the majority of differences with the reference is lower than ±10%. Variations in the compositions of the reference product have been observed to influence the results and a control criteria are proposed. Conclusion: The variability of the pharmacokinetic performance of teicoplanin can be better controlled with this new proposal of composition specification given by EMA.
Analytical specification, pharmacopeia, relative bioavailability, teicoplanin
Introduction The majority of glycopeptide antibiotics, from fermentative origin, consists of a combination of closely-related molecules and its resulting activity depends on the individual contribution of each active component. Concerning Teicoplanin, a Vancomycinlike antibiotic, the active subcomponents are identified as the ‘‘A2’’ group. All pharmacopeias describe the same five A2 subcomponents that are numbered based on a rank of retention time values. Concerning the chemical differences between each subcomponent, the molecule core is the same for all of them, but each one contains a slightly different aliphatic radical chain giving slight differences in polarity to the whole molecule1. Consequently, their specific pharmacokinetic characteristics are directly related with their physicochemical properties2. Although injectable generic drug applications to regulatory agencies are generally waived from pharmacokinetic bioequivalence studies3, in the case of teicoplanin there is a dearth of evidence with respect to the equivalent systemic exposure between different pharmacopeia-compliant batches4,5.
Address for correspondence: Dr. Antonio Boix-Montan˜es, Departamento de Farmacia y Tecnologia Farmace´utica, Facultad de Farmacia/ Universidad de Barcelona, Av. Joan XXIII, 08028 Barcelona, Spain. Tel: +34 934024560. Fax: +34 944024563. E-mail: [email protected]
History Received 9 December 2014 Revised 30 January 2015 Accepted 6 March 2015 Published online 17 April 2015
Assay specifications of this antibiotic in the European and Japanese pharmacopeias6,7 comprise the microbiological potency of the whole substance and also the chromatographic composition of A2 subcomponents. In this sense, the European Medicines Agency (EMA) has recently recommended8 to constrain the current specification of the chromatographic composition. This article examines how to find a scientific rationale to this modification using a practical case. Brink et al.9 pointed out that significant differences in A2i compositions (having a higher percentage of hydrophilic subcomponents) may lead to a lower sum of the areas under the teicoplanin plasma concentration versus time curve (AUC) of the A2i subcomponents. In addition, Rossi et al.10 have performed a principal component analysis (PCA) of a collection of A2i compositions of reference and generic batches with an ad hoc statistical model stating that this uncertainty between batches could only be solved by quantifying the effect on tissue drug concentrations. Certainly, it would be the best demonstration if an accurate knowledge of those levels would be achievable. In practice, a more suitable approach is needed to compare differences between marketed batches. A comparison based on known pharmacokinetic differences of each A2 subcomponent is a more reliable approach although this information is scarcely available in the literature. As an estimator of pharmacokinetic exposure, very few authors have published the AUC of the individual A2 subcomponents in relation with the
Pharmaceutical Development and Technology Downloaded from informahealthcare.com by Kainan University on 04/28/15. For personal use only.
2
A. Boix-Montan˜es & A. Garcia-Arieta
Pharm Dev Technol, Early Online: 1–4
respective HPLC composition of this injectable formulation1,11. Assuming this scarcity of useful data, we have related in a previous paper12 the experimental A2i compositions of a set of teicoplanin batches (normalized by the corresponding amount originally infused) with their respectives AUCs taken from the literature. In clinical practice, the AUC of teicoplanin is estimated as the linear combination of the contribution of each A2 subcomponent. In our case, the expected whole AUC of any teicoplanin batch has been estimated as the linear combination of these calculated relationships weighted by the respective composition of each batch of substance. As a result, some generic batches were compared based on this prediction of systemic exposure. The aim of this work is to try out this calculation approach to assess the pharmacokinetic significance of the specification proposed by EMA and also the relevance of the inter-batch differences of this drug using recently published sets of real data.
Methods A set of A2 compositions of 14 generic batches of teicoplanin that were published in the literature10 has been used to estimate their expected AUCi values using the individual response factors calculated as detailed in Ref. 12. A single AUC value has been calculated for each batch as the arithmetic sum of the contributions of each A2 component1. Individual AUCs have been calculated from the A2 composition using an AUC response factor of each subcomponent weighted by its respective percentage in the teicoplanin composition in each batch (AUCTEST). Two different sets of teicoplanin reference compositions have been comparatively used in this study to calculate a representative AUC value for the reference (AUCREF): (1) A pool of six batches from four different laboratories (24 A2 compositions)12. (2) A set of 11 batches of the reference product that were analyzed simultaneously with the 14 generic batches under study10. The corresponding mean values of each series were named as Ref1 and Ref2, respectively. Finally, the corresponding ratios AUCTEST/AUCREF for each generic batch have been calculated. Two series of results have been obtained using either Ref1 or Ref2 and their differences have been discussed. The obtained AUC ratios express the deviation of the AUC of the test batch (i.e. a generic batch) from the AUC given by a reference composition. In a second investigation, the estimations of AUCs have been used to estimate the inter-batch variability in A2 composition using the largest set of values available from a single manufacturer: Ref2. For this purpose, individual AUC values of Ref2 have been used as test values (numerator) and their mean has been used as the reference AUC (denominator).
Finally, the new specification proposed by EMA has been tested by calculating its corresponding hypothetical extreme compositions in A2 subcomponents. The theoretically lowest AUC would result from the maximum possible content of the most polar components (A2-1, A2-2, etc.). Meanwhile, the highest AUC would result from the maximum acceptable content of the most lipophilic ones (A2-5, A2-4, etc.). These ‘‘side-batches’’ have been taken to calculate the corresponding AUCTEST values of the extremes that are implicitly accepted by this specification. A summary of those values is shown in Table 1.
Results Mean AUC values (standard deviations within parentheses) for each reference data set are 325.71 (2.57) mg/mL h for Ref1 and 313.89 (8.50) mg/mL h for Ref2. Table 2 summarizes the estimated AUC values of the 14 test batches and resulting ratios using both Ref1 and Ref2 mean values. Using Ref1, the AUCs of 8 of 14 generic batches would differ more than ±5% of the reference and 3 of 14 batches would differ more than ±10%. Considering Ref2, 6 of 14 fall outside ±5% and 3 of 14 fall outside a ±10% level. In both cases, batch UKB10 turned out to be the worst case. It is shown that the equivalence of a generic batch with a reference composition depends not only on the test value but also on the specific value taken as the reference. Concerning the variability of the set of Ref2 batches, 3 of 11 batches fall outside a ±5% range and none of them outside the ±10% of their mean value. Resulting AUC ratios of each side-batch with the corresponding mean AUCREF range between 89.03% and 116.31% (from 10.97% to +16.31%) with Ref1 and between 92.39% and 120.71% (from 7.61% to +20.71%) with Ref2.
Table 2. Individual AUC values (mg/mL h) and their corresponding ratios for each generic batch considering references Ref1 and Ref2. Batch AB951 CO587 CO608 CR111 E275A FE004 GZ802 GZ805 K6802 LA103 LA104 LA105 UE821 UKB10
Estimated AUC
Ref1
Ref2
321.72 307.46 311.28 291.52 304.20 299.27 344.93 328.33 349.50 333.98 322.10 324.75 276.70 265.64
98.77 94.39 95.56 89.49 93.39 91.87 105.89 100.79 107.29 102.53 98.88 99.70 84.94 81.55
102.50 97.95 99.17 92.87 96.91 95.34 109.89 104.60 111.35 106.40 102.62 103.46 88.15 84.63
Bold numbers indicate deviations higher than ±5%.
Table 1. Specifications of A2 composition (%) of the European pharmacopeia (Reference 7) and of the EMA proposal (Reference 8) and their corresponding theoretical side-batches. Reference European pharmacopeia v. 7.0 Eur low side-batch Eur high side-batch EMA proposal EMA low side-batch EMA high side-batch
A2–1
A2–2
A2–3
A2–4
A2–5
Sum
0–20.0 20.0 0 10.0–19.0 19.0 10.0
35.0–55.0 55.0 40.0 37.0–50.0 46.0 40.0
0–20.0 5.0 20.0 5.0–11.0 5.0 11.0
0–20.0 0 20.0 7.0–15.0 7.0 15.0
0–20.0 0 20.0 7.0–17.0 7.0 17.0
80.0–100.0 80.0 100.0 84.0–93.0% 84.0 93.0
DOI: 10.3109/10837450.2015.1035726
Pharmaceutical Development and Technology Downloaded from informahealthcare.com by Kainan University on 04/28/15. For personal use only.
Discussion Normally distributed variations have been assumed for both sets of reference compositions. Considering that Ref1 includes the data from four different laboratories instead of one in the case of Ref2, inclusion of the interlaboratory variability results in a more representative estimation of the population of reference batches. We have assumed that Ref1 could be a better estimation of the true data. Using Ref1, the newly proposed EMA specification seems to imply the acceptance of maximum differences in AUC from around 10% to +15%. Meanwhile, the current European pharmacopeial specification leads to around a ±20% acceptation12, which is wider. Those results suggest that this new specification would improve the batch consistency. As expected, the modification of the specification of A2i composition suggested by EMA would control the therapeutic performance of teicoplanin better than the current pharmacopeial specification because the width of the acceptance range has been narrowed. Other proposals of A2 composition would be possible and could be assessed using this calculation approach based on the expected pharmacokinetic AUC. The noticeable variations of the reference value (Ref2) influence the result of the comparison. An approach to control this uncertainty could be to include a significant number (e.g. n 6) of A2 compositions of different reference batches in the same analytical run of the test batches and to accept the same variability that is accepted with the generics under testing, e.g. not more than ±10% departure from the mean value. Concerning the clinical relevance of the comparisons, it is known that principal components are defined as linear combinations of the original variables preserving the original variance13. The PCA of teicoplanin batches published by Rossi et al.10 searches for the maximum discrimination between reference and innovator batches to achieve a maximum separation among predefined classes. This actual combination means a feature selection instead of a feature reduction which should be the main goal of a PCA. In this sense, this mathematical approach is inappropriate to decide about clinical relevance of batch differences. In fact, drawing conclusions between batches should have to be based on biological considerations such as a pharmacokinetic performance or the expected antimicrobial efficacy of each batch. Nevertheless, a comparison based on MIC values has certain limitations as: MIC values found in the literature are discrete values (not continuous), data of the literature cannot be pooled (thorough information of analytical conditions is lacking), lower analytical selectivity than a chromatographic method, poorer predictability of this in vitro test compared with the in vivo pharmacokinetics, etc. Alternatively we have compared, with an agreed mean reference, all the individual values of this dataset of 14 generic batches on the basis of a pharmacokinetic descriptor of systemic exposure. As a result, the majority of AUC values of those batches differs less than ±5% of the reference composition, and few batches differ more than ±10%. Despite the scarcity of information in the literature about the relationship between A2 chemical compositions and its achievable plasma AUCs, it is possible to demonstrate a pharmacokinetically-based equivalence of a generic batch with a pool of well-standardized reference batches. It is also possible to predict the width of the acceptance range that corresponds to any A2 specification proposal. Differences among pharmacopeial compliant batches of other fermentative drugs have been also controversial regarding the microbiological or clinical comparability between different origins or manufacturers14,15, being attributed to differences in purification steps. If it was the case, further quality specifications should be proposed in addition to the chemical composition.
Equivalence between batches of a glycopeptide drug
3
It is worth noting that this article refers only to the specification of the chemical composition of teicoplanin and not to a conventional bioequivalence calculation. This comparison of the compositions of the active components based on bioavailability concepts has been shown to be useful to understand the modification of the current chemical specification values moving to more relevant ranges in terms of a better controlled pharmacokinetic exposure. In summary, this calculation approach has quantified the expected pharmacokinetic differences (described by AUC, the more suitable descriptor of bioavailability extent) due to inter-batch differences in chemical composition. It is remarkable to note that pharmacokinetic differences with the new specification would be lower than observed using current specifications. The acceptance range would be more restrictive and a more homogeneous systemic exposure of this antibiotic would be easily guaranteed.
Conclusions The prediction of the expected area under the plasma levels versus time curve of each active substance is a possible approach to control the risk of exposure variations of teicoplanin batches from different origins. This approach has been used to evaluate the pharmacokinetic performance of any specification of the chemical composition of A2 subcomponents and also to test the equivalence of different test batches with a pre-defined reference. This methodology can be applied to similar multicomponent antibiotics where a clinical rationale for the definition of its chemical composition is required.
Declaration of interest The authors report no conflicts of interest.
References 1. Bernareggi A, Danese A, Cometti A, et al. Pharmacokinetics of individual components of teicoplanin in man. J Pharmacokinet Biopharm 1990;18:525–543. 2. Rowland M. Clinical pharmacokinetics of teicoplanin. Clin Pharmacokinet 1990;18:184–209. 3. European Medicines Agency. Guideline on the investigation of bioequivalence. CPMP/EWP/QWP/1401/98 Rev. 1/Corr. 2010 Available from: http://www.emea.europa.eu/docs/en_GB/document_ library/Scientificguideline/2010/01/WC500070039.pdf [last accessed 13 Jun 2014]. 4. European Medicines Agency. Questions and answers on the referral for teicoplanin Hospira powder and solvent for injection containing 200 or 400 mg teicoplanin. EMA/CHMP/77992/2010. EMEA/H/A29/1084. 2009. Available from: http://www.ema.europa.eu/docs/ es_ES/document_library/Referrals_document/Teicoplanin_Hospira_ 29/WC500014116.pdf [last accessed 17 Feb 2013]. 5. Fujimura S, Fuse K, Takane H, et al. Antibacterial effects of brandname teicoplanin and generic products against clinical isolates of methicillin-resistant Staphylococcus aureus. J Infect Chemother 2011;17:30–33. 6. Japanese Ministry of Health, Labour and Welfare. Japanese Pharmacopoeia 16th ed. Teicoplanin monograph. 2012. Available from: http://jpdb.nihs.go.jp/jp16e/JP16.pdf [last accessed Nov 2012]. 7. European Council. European pharmacopoeia 7.0 corrected 6.6. Teicoplanin monograph (01/2009:2358). Available from: http:// www.edqm.eu/en/european-pharmacopoeia-publications-1401.html [last accessed Nov 2012]. 8. European Medicines Agency. Assessment report. Review under Article 5(3) of Regulation (EC) No 726/2004. Teicoplanin. Procedure no. EMEA/H/A-5(3)/1315. EMA; 21 March 2013. Available from: http://www.ema.europa.eu/docs/enGB/document library/Report/2013/04/WC500142229.pdf [last accessed 13 Jun 2014]. 9. Brink AJ, Richards GA, Colombo G, et al. Multicomponent antibiotic substances produced by fermentation, implications for
4
A. Boix-Montan˜es & A. Garcia-Arieta
Pharmaceutical Development and Technology Downloaded from informahealthcare.com by Kainan University on 04/28/15. For personal use only.
regulatory authorities, critically ill patients and Generics. Int J Antimicrobial Agents 2014;43:1–6. 10. Rossi A, Buttini F, Colombo P, et al. Complex product composition generates risks for generic substitution also with dosage forms for intravenous administration. Int J Pharm 2013;451:50–56. 11. Falcoz C, Ferry N, Pozet N, et al. Pharmacokinetics of teicoplanin in renal failure. Antimicrob Agents Chemother 1987;31:1255–1262. 12. Boix-Montan˜es A, Garcia-Arieta A. Composition specification of teicoplanin based on its estimated relative bioavailability. Drug Dev Ind Pharm 2015;41:218–223.
Pharm Dev Technol, Early Online: 1–4
13. Massart DL, Vandeginste BGM, Deming SN, et al. Chemometrics, a textbook. Data handling in science and technology. Amsterdam: Elsevier; 1988. 14. Vesga O, Agudelo M, Salazar BE, et al. Generic vancomycin products fail in vivo despite being pharmaceutical equivalents of the innovator. Antimicrob Agents Chemother 2010;54:3271–3279. 15. Agudelo M, Rodriguez CA, Pelaez CA, Vesga O. Even apparently insignificant chemical deviations among bioequivalent generic antibiotics can lead to therapeutic nonequivalence: the case of meropenem. Antimicrob Agents Chemother 2014;58:1005–1018.
