The Pharmacogenomics Journal (2013) 13, 452 -- 455 & 2013 Macmillan Publishers Limited All rights reserved 1470-269X/13 www.nature.com/tpj
ORIGINAL ARTICLE
Use of pharmacogenetics in bioequivalence studies to reduce sample size: an example with mirtazapine and CYP2D6 N Gonza´lez-Vacarezza1, F Abad-Santos2, A Carcas-Sansuan3, P Dorado1, E Pen˜as-LLedo´1, F Este´vez-Carrizo4 and A LLerena1 In bioequivalence studies, intra-individual variability (CVw) is critical in determining sample size. In particular, highly variable drugs may require enrolment of a greater number of subjects. We hypothesize that a strategy to reduce pharmacokinetic CVw, and hence sample size and costs, would be to include subjects with decreased metabolic enzyme capacity for the drug under study. Therefore, two mirtazapine studies, two-way, two-period crossover design (n ¼ 68) were re-analysed to calculate the total CVw and the CVws in three different CYP2D6 genotype groups (0, 1 and X2 active genes). The results showed that a 29.2 or 15.3% sample size reduction would have been possible if the recruitment had been of individuals carrying just 0 or 0 plus 1 CYP2D6 active genes, due to the lower CVw. This suggests that there may be a role for pharmacogenetics in the design of bioequivalence studies to reduce sample size and costs, thus introducing a new paradigm for the biopharmaceutical evaluation of drug products. The Pharmacogenomics Journal (2013) 13, 452--455; doi:10.1038/tpj.2012.29; published online 26 June 2012 Keywords: CYP2D6; genotypes; intra-individual; variability; bioequivalence; cost
INTRODUCTION In bioequivalence studies of highly variable drugs, intra-individual variability (CVw) is usually much greater than in other studies, which makes it necessary to substantially increase the number of subjects so as to attain an adequate statistical power. Consistently, a conventional two-way, two-period crossover study with 24 subjects has been found that would not be sufficient to demonstrate bioequivalence.1 Then, to be able to compare the bioavailable dose fraction between two drug products (reference and test) with reasonable accuracy and precision, it is important that the subjects’ clearance remains steady through both periods (Figure 1). In other words, subjects with high CVw in drug bioavailability, could show differences in drug plasma concentrations between drug formulations due to their inherent betweenperiod changes in metabolism, and not fully related to differences in the absorption of formulations under comparison.2 Thus, it can be expected that the inclusion of subjects with reduced metabolic enzyme capacity for the drug under study might decrease pharmacokinetic CVw and hence sample size and costs. About 25% of the most frequently prescribed drugs are mainly metabolised by CYP2D6 such as mirtazapine. The CYP2D6 gene that codes for CYP2D6 is highly polymorphic, with alleles producing absent catalytic activity.3 In all, 5--10% of Caucasian subjects carry just CYP2D6 inactive gene copies and therefore have a genetically determined null enzyme capacity 4,5 and about 30% may present just one active copy related to decreased enzyme capacity. As a consequence, drugs with relevant CYP2D6 metabolism, such as mirtazapine, appear as interesting probes to assess the potential effect of genotypes on CVw. Mirtazapine, a noradrenergic and specific serotonergic antidepressant, is a CYP2D6 and CYP3A4 substrate with an absolute
bioavailability of 50%, mainly caused by a relevant first-pass metabolism. CYP2D6 contributes to about 25% of total clearance of mirtazapine in subjects with only one active allele and up to 55% in the genetically defined ultrarapid metabolizers.6 Also, it is known that genetic polymorphisms influence pharmacokinetic parameters of ( þ )-mirtazapine, with an increase of 79% in the area under the concentration--time curve (AUC) for poor metabolizers (PMs; subjects with 0 CYP2D6 active genes) versus extensive metabolizers (subjects with one or more active genes).7 To the best of our knowledge, there has just been a bioequivalence study of tacrolimus showing that CVw differed across CYP3A5 genotypes.8 Therefore, we re-analysed two studies of mirtazapine to evaluate for the first time whether there were differences between the total CVw and the CVws according to different CYP2D6 genotypes (0, 1 and X2 active genes).9
MATERIALS AND METHODS Detailed information of the methods used in these two studies has been described elsewhere.9 A summary of the methods is presented below.
Study design Bioequivalence studies were carried out in a standard two-period, twosequence, crossover randomized design, executed by two centres in Madrid (Spain): the ‘La Paz’ and ‘La Princesa’ University Hospitals. Subjects received a single 30 mg oral dose of each mirtazapine formulation (Test and Reference) in each study period, with a washout period of at least 20 days. The Reference formulation used in both Centres was Rexer, Laboratorios Organon (Madrid, Spain). Test formulations were from Laboratorios Normon SA (Madrid, Spain) and from Laboratorios Alter SA
1 CICAB Clinical Research Centre, Extremadura University Hospital and Medical School, Badajoz, Spain; 2Service of Clinical Pharmacology, Hospital Universitario de la Princesa, Instituto Teo´filo Hernando, Instituto de Investigacio´n Sanitaria Princesa (IP), Madrid, Spain; 3Clinical Pharmacology Service, Hospital Universitario La Paz, Pharmacology Department, School of Medicine, Universidad Auto´noma de Madrid, Madrid, Spain and 4Centre for Biomedical Sciences, University of Montevideo, Montevideo, Uruguay. Correspondence: Dr A Llerena, CICAB, Clinical Research Centre, Extremadura University Hospital and Medical School, Badajoz 06080, Spain. E-mail:
[email protected] Received 27 February 2012; revised 10 May 2012; accepted 23 May 2012; published online 26 June 2012
Use of pharmacogenetics in bioequivalence studies N Gonza´lez-Vacarezza et al
AUCT AUCR
=
FT × D ClT
453 ×
ClR FR × D
≅
CVW (%) = 100 × e MSE −1
FT
2 ⎛ CVW ⎞ × (t 2 ⎟ N ≥ 2×⎜ / 2 , N − 2 + t ,N − 2 ) (II) ⎝ 0.2 ⎠
FR
Figure 1. AUC ratio for oral administration. AUC, area under the concentration-time curve; F, bioavailable dose fraction; D, dose; Cl, clearance; T, Test drug product; R, Reference drug product.
(I)
Figure 2. (I) Formula for CVw (%) determination. MSE, mean squared error. (II) Formula for sample size estimation in bioequivalence studies.
(Madrid, Spain) for the ‘La Paz’ Hospital and ‘La Princesa’ Hospital, respectively.
Subjects In all, 72 healthy volunteers, 36 in each study with a gender ratio of 1:1, were included. All participants were non-smokers, tested negative in a drugs-of-abuse urine screen, were not taking any concomitant medication, and were considered healthy after a physical examination, a 12-lead electrocardiogram, urinalysis, haematological and blood chemistry analysis, and a review of their clinical history. Body mass index values were between 18 and 30. Participants who tested positive for pregnancy were excluded from the study. The protocols were approved by the Ethics Committees for Clinical Investigation of the ‘La Paz’ and ‘La Princesa’ Hospitals.
Study protocol Participants were not allowed to consume alcohol, caffeine, chocolate, tea, or cola-type beverages for at least 24 h before each mirtazapine dose, and were not allowed to use any other drug starting 2 days before the beginning of the study. Participants fasted from 10 h before, until 5 h after the administration of the medication, at which time a standard lunch was served. Venous blood samples were collected from each volunteer during each of the study periods from 0 to 96 h or 120 h after drug administration. Physical examination, a 12-lead electrocardiogram, urinalysis, and a haematological and blood chemistry analysis were performed not only before inclusion in the study, but also 2 weeks after the second period. Analytical method. Plasma concentrations of mirtazapine were determined by high-performance liquid chromatography with a coupled mass spectrometry-validated method, in compliance with good laboratory practices. Genotyping method. Genomic DNA was isolated from whole blood from each subject using the Puregene DNA Isolation Kit (Gentra Systems, Minneapolis, MN, USA). The DNA was quantified spectrophotometrically and stored at 4 1C. The CYP2D6 (*3, *4, *6, *7 and *9) alleles were determined by allele-specific PCR. The alleles with deletion (*5) and duplication of CYP2D6 gene were analysed by long PCR. In accordance with standard procedures, the CYP2D6*1 variant allele was assigned as the one lacking the mutations analysed in the CYP2D6 gene. CYP2D6 genotyping was only available for 68 of the 72 subjects.8
Pharmacokinetic and statistical analyses Peak plasma concentration (Cmax) and time to maximum plasma concentration (Tmax) were determined directly from raw data. The AUC0--t was calculated by the trapezoidal rule, and AUC0--N was calculated as AUC0--t þ C(last)/lz, with C(last) being the last-measured concentration and lz the slope obtained from least-square regression of the terminal elimination phase. An ANOVA was performed on log-transformed data of the pharmacokinetic parameters of AUC0last, AUC0N and Cmax using SPSS- version 15.0 (SPSS, Chicago, IL USA). The ANOVA model included product, sequence, and period as fixed effects, and subject nested within sequence as random effect. The value of CVw was calculated using formula I in Figure 2. Although this formula does not exactly determine CVw, which might be estimated from a study with a replicate administration of the same formulation, it is the appropriate variance term for computing confidence intervals for the difference between formulations means.10 The sample sizes for the bioequivalence study were estimated considering a multi& 2013 Macmillan Publishers Limited
Table 1.
CYP2D6 genotypes and classification according to active
genes
Genotype
Group
CYP2D6 active genes
N
(*1/*1) 2 *1/*1 (*1/*4) 2 *1/*9 *1/*4 *1/*5 *1/*6 *4/*9 *3/*9 *4/*6 *4/*4
(II) (II) (II) (II) (I) (I) (I) (I) (I) (0) (0)
42 2 2 2 1 1 1 1 1 0 0
1 32 1 1 18 2 4 1 1 1 6
Abbreviations: (II), two or more active genes; (I), one active gene; (0), no active gene.
plicative model with a power of 80%, a significance level of 0.05, and a bioequivalence range (0.80--1.25) using formula II in Figure 2. Statistical analyses were performed to analyse the pharmacokinetic parameters and the estimation of sample sizes for the genotyped subjects. These were distributed into three groups according to their number of CYP2D6 active genes: two or more active genes ((II)), one active gene ((I)), no active gene ((0)). Additionally, analyses were also performed for two additional, created groups: one consisting of the individuals with 0 or 1 active gene ((0) þ (I)), and the other of those with 1, 2, or more than 2 active genes ((I) þ (II)). The group (I) þ (II) was considered because of the bimodal distribution of the phenotype groups for CYP2D6 in the Caucasian population.11 The CYP2D6 *3, *4, *5, *6 and *7 variants were considered as nonfunctional, whereas CYP2D6 *9 was considered active although it is a variant with reduced activity (Table 1).
RESULTS Genotypes were determined for 68 of the 72 subjects (36 women and 36 men) included in the studies. The CYP2D6 genetic variants and their classification into three groups based on the number of CYP2D6 active genes are detailed in Table 1. Table 2 gives the geometric mean, mean squared error and CVw for each group. In group (II) (n ¼ 35), the CVw of AUC0last, AUC0N and Cmax were 95.1%, 78.2% and 16.4% greater, respectively, than in group (0) (n ¼ 7). However, groups (I) (n ¼ 26) and (0) had similar CVw values. The observed CVw differences may influence sample size estimation for bioequivalence studies (Table 3). As Cmax is a limiting factor for determining the sample size, the number of subjects could be reduced by 29.2% by recruiting only PM individuals. Similarly, a reduction of 8.3 or 15.3% could be achieved by including individuals from groups (I) or (0) þ (I), respectively, as against the total group. On the contrary, taking individuals in groups (I) þ (II) or (II) would seem to lead to the opposite tendency, with increases in sample size of 5.6% and 12.5%, respectively (Table 3). The Pharmacogenomics Journal (2013), 452 -- 455
Use of pharmacogenetics in bioequivalence studies N Gonza´lez-Vacarezza et al
454 Table 2. Analysis of pharmacokinetic parameters for the three groups separately, according to the number of CYP2D6 active genes ((0), (I) and (II)), for the groups including individuals in groups (0)+(I) and (I)+(II) and for the total of individuals regardless of CYP2D6 genotype (0) (N ¼ 7)
(I) (N ¼ 26)
(II) (N ¼ 35)
(0)+(I) (N ¼ 33)
(I)+(II) (N ¼ 61)
Total (N ¼ 68)
LnAUC0last Geometric mean (ng h ml1) MSE CVw(%) CI 95% %a
899.2 0.00589 7.69 [4.79--18.99] 0
757.8 0.00711 8.45 [6.59--11.77] 9.9
669.4 0.0222 15.0 [12.1--19.8] 95.1
785.8 0.00655 8.10 [6.50--10.79] 5.3
705.8 0.0163 12.8 [10.9--15.7] 66.4
723.6 0.0150 12.3 [10.5--14.8] 59.9
LnAUC0N Geometric mean (ng h ml1) MSE CVw(%) CI 95% %a
947.5 0.00696 8.36 [5.21--20.68] 0
792.8 0.00758 8.72 [6.81--12.16] 4.3
702.5 0.0219 14.9 [12.0--19.7] 78.2
823.3 0.00703 8.40 [6.73--11.18] 0.48
739.6 0.0174 13.2 [11.2--16.2] 57.9
758.7 0.0150 12.3 [10.5--14.8] 47.1
LnCmax Geometric mean (ng ml1) MSE CVw(%) CI 95% %a
80.1 0.0447 21.4 [13.3--55.6] 0
81.1 0.0481 22.2 [17.3--31.2] 3.7
70.4 0.0602 24.9 [20.0--33.2] 16.4
80.8 0.0450 21.5 [17.1--28.8] 0.47
74.7 0.0554 23.9 [20.2--29.3] 11.7
75.3 0.0532 23.4 [19.9--28.4] 9.3
Abbreviations: CI 95%, 95% confidence interval for CVw; Cmax, maximum concentration; CVw (%), intra-subject coefficient of variation; (0), subjects with no CYP2D6 active gene; (I), subjects with one CYP2D6 active gene; (II), subjects with two or more CYP2D6 active genes; LnAUC0last, log-transformed area under the concentration-time curve from 0 to the last quantifiable concentration; LnAUC0N, log-transformed AUC0N; MSE, mean squared error. a Percentage increment with respect to group (0) CVw(%).
Table 3.
Estimation of sample size of bioequivalence studies for the three groups separately, according to number of CYP2D6 active genes ((0), (I) and (II)), for the groups including individuals in groups (0)+(I) and (I)+(II) and for the total of individuals regardless of CYP2D6 genotype
LnAUC0last LnAUC0N LnCmax
T/R
(0)
(I)
(II)
(0)+(I)
(I)+(II)
Total
1.05 1.10 1.05 1.10 1.05 1.10
4 9 4 10 23 51
4 9 4 9 30 66
13 30 13 29 36 81
4 9 4 9 27 61
10 22 10 23 34 76
9 20 9 20 32 72
Abbreviations: Cmax, maximum concentration; EM, extensive metabolizer subjects (I)+(II); LnAUC0last, log-transformed area under the concentrationtime curve from 0 to the last quantifiable concentration; LnAUC0N, logtransformed AUC0N. Sample size estimated with the formula II indicated in Figure 2, for Test:Reference ratios of 1.05 (or 0.95) and 1.10 (or 0.90). (0), poor metabolizer subjects; (I), subjects with one active gene; (II), subjects with two active genes.
There were no statistically significant differences between groups in body mass index (P ¼ 0.064), sex (P ¼ 0.305), age (P ¼ 0.402), or sequence assignments (P ¼ 0.588).
DISCUSSION Subjects with absent CYP2D6 activity (PMs or group (0)) have lower pharmacokinetic CVw than groups (I) þ (II), (II) and the total group. However, no statistically significant differences were found due to the low number of individuals in this group. PMs would have less drug pharmacokinetic variability between periods, due to the lack of elimination via the CYP2D6 metabolic pathway. Furthermore, this group more accurately reflects differences between the bioavailabilities of drug products, due to the lesser The Pharmacogenomics Journal (2013), 452 -- 455
inter-period variability in their own metabolism. In the present study of mirtazapine, recruiting just PMs would give a reduction of about 29.2% in the number of subjects required. The present data are given as an example of how it might be possible to reduce the pharmacokinetic CVw, and consequently sample size and costs, by including subjects with decreased metabolic enzyme capacity. For instance, if one assumes an average total cost of the study per participant of 4000 euros, and theoretical geometric mean ratios of 1.05 and 1.10, then the inclusion of only CYP2D6 PMs would save up to 36 000 and 84 000 euros, respectively. Similarly, the inclusion of individuals with 0 and 1 active CYP2D6 gene would save up to 20 000 and 44 000 euros, respectively. Nevertheless, the most important limitation to implementing such a strategy based on the study of CYP2D6 PMs is the low prevalence of this group. Therefore, according to the present data, it would be possible for mirtazapine studies to also recruit subjects with one active copy ((0) þ (I) group), because groups (0) and (I) have similar values of CVw, and show lower variability in pharmacokinetic parameters over time than group (II). Regulatory agencies indicate that bioequivalence studies must be conducted in homogeneous but representative samples of the general population, ensuring external validity of the results. This is clearly essential for the analysis of drug safety and efficacy to ensure that the results can be generalized to the whole population. However, as the main purpose of bioequivalence studies is to assess differences among drug products’ bioavailabilities, decreasing intra-individual variability might increase the accuracy of determinations of drug-related factors.
CONFLICT OF INTEREST The authors declare no conflict of interest.
ACKNOWLEDGEMENTS This work was partially supported by a scholarship of the National Agency for Research and Innovation of Uruguay, by the Instituto de Salud Carlos III, and EU
& 2013 Macmillan Publishers Limited
Use of pharmacogenetics in bioequivalence studies N Gonza´lez-Vacarezza et al
455 FEDER Grants PI10/02758 and CP06/00030 (PD). The study was coordinated in the networks CIBERSAM and CAIBER, which are initiatives of ISCIII (Spain).
REFERENCES 1 Davit BM, Conner DP, Fabian-Fritsch B, Haidar SH, Jiang X, Patel DT et al. Highly variable drugs: observations from bioequivalence data submitted to the FDA for new generic drug applications. AAPS J 2008; 10: 148--157. 2 Vaan Peer A. Variability and impact on design of bioequivalence studies. Basic Clin Pharmacol Toxicol 2010; 106: 146--153. 3 CYP2D6 allele Nomenclature. http://www.cypalleles.ki.se/cyp2d6.htm (Accessed 17 June 2011). 4 Llerena A, Cobaleda J, Martı´nez C, Benı´tez J. Interethnic differences in drug metabolism: influence of sex-related and environmental factors on debrisoquine hydroxylation phenotype. Eur J Drug Metab Pharmacokinet 1996; 21: 129--138. 5 Dorado P, Pen˜as-Lledo´ EM, Llerena A. CYP2D6 polymorphism: implications for antipsychotic drug response, schizophrenia and personality traits. Pharmacogenomics 2007; 8: 1597--1608.
& 2013 Macmillan Publishers Limited
6 Kirchheiner J, Henckel HB, Meineke I, Roots I, Brockmo¨ller J. Impact of the CYP2D6 ultrarapid metabolizer genotype on mirtazapine pharmacokinetics and adverse events in healthy volunteers. J Clin Psychopharmacol 2004; 26: 647--652. 7 Timmer CJ, Sitsen JM, Delbressine LP. Clinical pharmacokinetics of mirtazapine. Clin Pharmacokinet 2000; 38: 461--474. 8 Yong Chung J, Jung Lee Y, Bok Jang S, Ahyoung Lim L, Soo Park M, Hwan Kim K. CYP3A5*3 genotype associated with intra-subject pharmacokinetic variation toward tacrolimus in bioequivalence study. Ther Drug Monit 2010; 32: 67--72. 9 Borobia AM, Novalbos J, Guerra-Lo´pez P, Lo´pez-Rodrı´guez R, Tabares B, Rodrı´guez V et al. Influence of sex and CYP2D6 genotype on mirtazapine disposition, evaluated in Spanish healthy volunteers. Pharmacol Res 2009; 59: 393--398. 10 Chen ML, Lee SC, Ng MJ, Schuirmann DJ, Lesko LJ, Williams RL. Pharmacokinetic analysis of bioequivalence trials: implications for sex-related issues in clinical pharmacology and biopharmaceutics. Clin Pharmacol Ther 2000; 68: 510--521. 11 LLerena A, Dorado P, Pen˜as-Lledo´ EM. Pharmacogenetics of debrisoquine and its use as a marker for CYP2D6 hydroxilation capacity. Pharmacogenomics 2009; 10: 17--28.
The Pharmacogenomics Journal (2013), 452 -- 455