Pharmacogenetic Perspectives in Improving Pharmacokinetic Profiles for Efficient Bioequivalence Trials with Highly Variable Drugs: A Review

Abstract

Conducting bioequivalence trials under traditional crossover study designs without exposing a large number of healthy volunteers to demonstrate two highly variable (%coefficient of variability greater than 30) test/reference (branded) drug products in different formulations to meet the standard 90% confidence interval criteria of relevant pharmacokinetic metrics between 0.80 and 1.25 and to maintain the consumer risk smaller than 5% has been a challenging task. Genetic polymorphisms encoding key drug-metabolizing enzymes can significantly influence absorption, distribution, metabolism and elimination of many highly variable generic drugs after administration. This article briefly reviews the case studies and examples of utilizing pharmacogenetic screening approaches in the recent literature to alleviate the resources and ethical burden of recruiting larger numbers of subjects in bioequivalence trials needed to perform pharmacokinetic studies for formulations of highly variable drug products without widening the bioequivalence acceptance limits.

Keywords::

• bioequivalence
• generics
• genetic polymorphisms
• highly variable drugs
• intra-individual variability
• pharmacogenetic
• pharmacokinetics

The innovative pharmaceutical companies are granted exclusive marketing rights for carrying a new medicine for a certain time period after approval by regulatory agencies. After a limited period is concluded, other drug companies could file an abbreviated new drug application for generic approval to generate their own generic versions via bioequivalence studies where pharmacodynamics/pharmacokinetic related metrics of the test/reference (branded) formulations such as area under the concentration (AUC)-time curve and peak plasma concentration (C_max) after drug administration are expected to follow a 90% CI for the average ratios of the systematic exposure measures to fall within the constraint of 80–125% acceptance limits [1–3]. Commonly, standard bioequivalence trials are carried out in healthy volunteers as a two-way crossover, randomized design for subjects to take both test and reference (branded) formulations in consequent periods. For generic drug products with intrasubject variability smaller than 30%, the difference between formulations was basically the determinant of the fulfillment of bioequivalence criteria, but for those with increasing intrasubject coefficient of variability (ICV) over 30%, ICV instead of formulation factor became decisive [4]. The larger ICV leads to the wider estimated CI, and it becomes very difficult to remain within the predetermined bioequivalence limits, despite the fact that an approved medicine with higher intrasubject variability characteristic normally possess relatively wider therapeutic index. As indicated in Figure 1, examples exist where a highly variable test product failed to demonstrate bioequivalence when compared with itself in a bioequivalence study using the standard design/sample size [5]. Therefore, variability is one of the fundamental detriments of bioequivalence studies on determining whether two dosage forms of the same chemical compound behave interchangeable in clinical practice. For example, if the intra-individual variability values for AUC and C_max measured from a repeat study of the reference formulation of progesterone, a female hormone for the regulation of ovulation and menstruation, from a few postmenopausal females are over 60 and 95%, respectively, then a generic company might require dosing in projected 300 postmenopausal women to achieve adequate statistical power for a standard two-period crossover bioequivalence study.

Figure 1. Representative results of the statistical analyses of bioequivalence studies.

The two bars represent the widths of hypothetical 90% confidence intervals from bioequivalence studies of generics with regular variability (green bar) and high variability (red bar). The red bar represents the 90% confidence intervals of the bioequivalence study C_max or AUC test/reference ratios normally distributed about the point estimate. In this illustration, the 90% CI of the highly variable generics can exceed bioequivalence limits solely because of the intrinsic variability. Statistically using more subjects in the bioequivalence study will cause the 90% confidence interval of a highly variable drug to become narrower to fall within the bioequivalence acceptance limits of 80–125%.

AUC: Area under the concentration; C_max: Peak plasma concentration.

The ICV in the human pharmacokinetics of a dosed drug can be inherently due to the nature of drug, the formulation factors of the products, bioanalytical techniques and interaction with other physiologic and pathophysiologic factors such as administration conditions, transporter, first-pass metabolism and so on [6–16]. A feasible remedy is to increase the number of study participants in a study and thereby to narrow the CI. However, the development for high variable (HV) generics will become very expensive and cumbersome. Except for recruiting large numbers of subjects [17–22], other possible solutions such as using replicating, group sequential study designs [23–29] and multiple dosing with nonradioactive isotopes [30], have been proposed to deal with the problems of generics possessing larger root mean square error values of relevant pharmacokinetic metrics in formulations meeting the bioequivalence limits. Although these approaches take a longer time to complete a replicate design study (with the potential of increasing subject dropout rate from the trials), the sample size required for a replicate design study can be reduced up to 50% since study subjects are each employed for twice as many periods.

Davit et al. of the US FDA [5] had reviewed 1010 bioequivalence studies of 180 drugs in 2003–2005 submissions, of which around 31% (57 out of 180) of the surveyed drugs were highly variable and 10%, 39% and 51% of these highly variable drugs were classified either borderline, inconsistently or consistently highly variable, respectively, as a result of the pharmacokinetic characteristics of the drug substances. Analysis of the data revealed that extensive first-pass metabolism variability was the most important factor over food effect and physicochemical characteristics contributing to high variability on top of formulation performance, but also influencing the different degree of variability on both AUC and C_max in bioequivalence trials. It is well known that pharmacogenetic variation within an individual could have clinical consequences modulating protein function and hence drug metabolism through multiple mechanisms. Genetic factors seem to account widely for the activity variation of drug-metabolizing enzymes among healthy people, potentially making drug metabolism on critical pharmacokinetic end points of test or reference formulations highly variable. This manuscript reviews proposals and case examples published in the literature using pharmacogenetic-based methodologies for the efficient bioequivalence evaluation of generic HV drug products.

Materials & methods

To review the practicability of pharmacogenetic methods to clinical implementation, the available information from the literature to establish efficient bioequivalence studies of HV drugs on formulations, given the difficulty for them to meet the standard regulatory acceptance limit without increasing human experimentation (number of volunteers) and excessive consequential trials, is summarized in Table 1, which was searched through PubMed covering the period from 1 January 1994 to 1 December 2019. Throughout this section, the fundamental pharmacogenetic–pharmacokinetic techniques used for determining if HV test/reference drug products are interchangeable with minimum number of study subjects are described below.

Table 1. Summary of bioequivalence trials for highly variable drugs using pharmacogenetic screen.

Study (year)	n	Drug	Genotypes	Pharma-cokinetic and statistical analyses	Study design	Polymorphism-altering pharmacokinetic and bioequivalence outcomes	Ref.
LLerena (2013), Spain	72	Mirtazapine	CYP2D6	Cmax AUC0-120 h AUC0-∞ ANOVA	30 mg two-period cross-over studies with a 20-day washout period re-analyzed in three different genotypes	Around1/3 reduction in the minimum number of subjects could be achieved by recruiting poor metabolizers Polymorphisms on AUC variability is less importance than Cmax in bioequivalence	[31]
Abad-Santos (2015), Spain	70	Risperidone	CYP2D6	Cmax AUC0-24 h ANOVA	1 mg two-period cross-over studies with a 14-day washout period were reanalyzed and subjects were classified into four different phenotype groups	Gender yielded less effect than genetic polymorphisms on within-subject variability. Genotyping for clinically relevant polymorphisms was recommended for subjects participating in bioequivalence	[32]
Chen (2018), China	18	Risperidone	CYP2D6	Cmax AUC0-96 h AUC0-∞ ANOVA	4 mg standard 2 × 2 cross-over with a 2-week washout period	By controlling the genotype to reduce intra-individual variability, the test formulation met regulation criteria for bioequivalence	[33]
Cho (2006) Korea	24	Risperidone	CYP2D6	Cmax AUC0-∞ ANOVA	2 mg two-period crossover with 1-week washout period	Two preparations interchangeable under clinical practice on the basis of genotypes	[34]
Hua (2018), China	260	Gefitinib	CYP3A4 CYP2D6	Cmax AUC0-216 h AUC0-168 h AUC0-∞ ANOVA	250 mg 2 × 2 cross-over with a washout period of 21 days	The activity of metabolizers was influenced with drug exposures but not bioequivalence results	[35]
Cedillo-Carvallo (2019), Mexico	70	Tolterodine	CYP2D6	AUC0-t ANOVA	2 mg 2 × 2 cross-over design with at least a 3‐day washout period	A minimum 30% reduction in sample size achieved by genotype-based enrichment strategy	[36]
Cedillo-Carvallo (2014), Mexico	36	Clopidogrel	CYP2C19	Cmax AUC0-36 h AUC0-∞ ANOVA	75 mg 2 × 2 cross-over with a 7-day washout period	Selected subjects homozygous for a certain haplotype increased the stringency of bioequivalence statistics	[37]
Chung (2015) Korea	104	Voriconazo-le	CYP2C19	Cmax AUC0-144 h AUC0-∞ ANOVA	200 mg 2 × 2 cross-over conducted after a single intravenous infusion and genotype identified and reanalyzed	Genotyping affected both pharmacokinetics and intra-individual variability	[38]
Jiang (2013) China	24	Citalopram	CYP2C19	Cmax AUC0-144 h AUC0-∞ ANOVA	20 mg, 2 × 2 cross-over with a 2-week washout period	The test and reference tablets met the regulatory bioequivalence criteria in the selected healthy male extensive metabolizers	[39]
Yong Chung (2010) S. Korea	29	Tacrolimus	CYP3A5	Cmax AUC0-96 h AUC0-∞ ANOVA	1 mg, 2 × 2 cross-over with a 21-day washout period	The sample size needed for specific genotypes to show bioequivalency can be reduced from nongenotyping cases	[40]
Choi (2015) Korea	46	Erlotinib	CYP1A1 CYP1A2 CYP3A4	Cmax AUC0-96 h ANOVA	150 mg, 2 × 2 cross-over with a 2-week washout period	CYP1A2 alone significantly affected metabolism; no corresponding pharmacokinetic variability reported	[41]
Peiró (2009) Spain	18	Tenoxicam	CYP2C8 CYP2C9	Cmax AUC0-∞ ANOVA	2 × 2 cross-over with a 3-week washout period	Phenotyping resulted in a reduced pharmacokinetic variability	[42]

ANOVA: Analysis of variance; AUC: Area under the concentration; C_max: Peak plasma concentration; n: Sample size.

Study design

Bioequivalence studies were carried out in a standard randomized, open-label, two-treatment, two-period, two-sequence, crossover study with a washout period of half-life-related time frames and executed by multiple centers in Hospitals. In each period, subjects randomly received a single oral or an intravenous infusion dose of a test or a reference drug formulation obtained from different sponsors. Serial blood samples of a certain volume for bioanalytical assays were taken at predose, and more than 12 certain sampling time points post dose.

Subjects & study protocol

Eligible healthy volunteers with both genders, no pregnancy and no any concomitant medication were included in the case studies and their weight, height, body mass index, physical examination, vital signs, laboratory tests and 12-lead ECGs were carefully performed and recorded before inclusion in the trials but also after the second period. Clinical history and adverse events were collected during the entire study period. Subjects using any drug known to induce or inhibit cytochrome P450 within 4 weeks prior to drug administration were excluded. All recruited subjects were provided written informed consent to become a volunteer participant in studies followed by an interpretation of study procedures. The protocols were approved by each institutional review board. All volunteers were aware of the risks and signed a clinical investigation agreement to participate in these studies.

Bioanalytical method

Concentrations of the dosed compounds in biological fluids were quantified by validated high-performance liquid chromatography coupled with mass spectrometry-based methods in compliance with good laboratory practices. All biological samples were pretreated prior to injection into analytical instruments.

Genotyping method

Clinically relevant polymorphisms among drug-metabolizing enzymes including various allelic variants that might influence intrasubject variations in the pharmacokinetic profile of individual generics in terms of efficacy and safety profile were genotyped using the TaqMan allelic discrimination assays. Predose blood samples for genotype assays were collected in K2EDTA anticoagulation tubes. Genomic DNA was extracted from peripheral blood in anticoagulation tubes and stored at −80°C until PCR analysis for genotyping. The DNA was quantified by spectrophotometry using validated genotyping assays. The CYP alleles were determined by allele-specific real-time PCR instrument. Subjects with a certain allelic variant were classified into specific genotype.

Pharmacokinetic & statistical analyses

C_max was determined directly from each observed individual time–concentration profile. Other pharmacokinetics parameters using a noncompartmental method and bioequivalence statistics were also performed with the commercial WinNonLin^® software package. Analysis of variance was performed on log-transformed data to compare the AUC_0-t, AUC_0-∞ and C_max using the factors fitted for the fixed effects of sequence, subject within sequence, period, treatment and subject nested within sequence as random effect. CYP genotypes were compared by the Χ² test. Results of the geometric least-square means and the 90% CI for AUC and C_max test/branded geometric mean ratio were presented for comparison. Bioequivalence was declared if the 90% CI of the treatment ratio was within the acceptable equivalence range of 0.8–1.25. The p < 0.05 was considered statistically significant.

Evaluation of intra-individual variability & estimation of sample size

The intra-individual variability (CV _w ) was calculated using the following formula:${\text{CV}}_{\text{ω}}=\sqrt{\left(e^{\text{MSE}}\right)-1}$

Where mean squared error (MSE) is the MSE obtained from analysis of variance or estimated from the study with replicate administration of the same formulation providing the appropriate variance term for computing CIs for the difference between genotypes and formulation.

The estimated sample sizes for the bioequivalence study were calculated using the formula based on a multiplicative model with a power of 80%, a significance level of 0.05 and a bioequivalence range of 0.80–1.25, as follows:$\text{N}\ge 2\times {\left(\frac{{\text{CV}}_{\text{ω}}}{0.2}\right)}^2\times {\left(t_{\frac{\text{α}}{2},\text{N}-2}+t_{\text{β},\text{N}-2}\right)}^2$

where N is sample number, N-2 the degree of freedom, t the t-test value and α and β statistical errors.

Application of pharmacogenetics in bioequivalence studies

The Cytochrome P450 (CYP) 1, CYP2 and CYP3 families are well-known most frequently metabolizing enzymes for small molecule drugs and a major source of variability in drug pharmacokinetics and clinical pharmacology. Their enzyme activities dependent on genetic polymorphisms and changes in physiological conditions might contribute to individual variations in in vivo pharmacokinetic/pharmacodynamics outcomes [43–59]. Expression of individual CYP families is well known to be associated with several factors including genetic polymorphisms, induction by xenobiotics, regulation by cytokines, hormones, gender, age and so on [43]. Multiallelic genetic polymorphisms could be generally classified to distinct pharmacogenetic phenotypes as poor, intermediate and extensive metabolizers for pharmacogenetic screening [43]. This review focuses on the applications of genetic polymorphisms screen of the CYP enzymes as an alternative approach in altering intra-individual variability of the pharmacokinetic profiles of HV drug/products conducted in bioequivalence studies. Table 1 provides an overview of recent case studies exploring pharmacogenetic selection on the CYP enzymes and their polymorphisms for bioequivalence evaluation of generic and branded formulations.

Case studies: CYP2D6

Mirtazapine, an antidepressant, is mediated by CYP2D6 contributing about 35% to total drug biotransformation with an absolute bioavailability of 50%. LLerena and coworkers in Spain [31] carried out a standard bioequivalence study in 72 participants with both sexes receiving a single 30 mg oral dose of each mirtazapine formulation (test vs reference products). CYP2D6 genetic variants were determined for most of volunteers included in the studies and classified into three groups (0, n = 7, no active gene, poor metabolizers), (I, n = 26, one active gene, intermediate metabolizers) and (II, n = 35, two or more active genes, extensive metabolizers). It was observed that in group (II), the intra-individual variability of AUC_0-t, AUC_0-∞ and C_max were 95.1%, 78.2% and 16.4%, respectively, greater than those in group (0) where groups (0) and (I) had comparable variation. Re-estimation of sample size of bioequivalence studies for all groups calculated through their pharmacokinetic variation data suggested that a 29 or 15% sample size reduction would have been achieved if the recruitment had been of individuals carrying just group 0 or a combination of groups 0 and 1, respectively, confirming the potential role of adopting pharmacogenetic screen in a bioequivalence study to reduce sample size and costs for drug development.

Risperidone, a derivative of benzisoxazole, metabolized by CYP2D6 is an antipsychotic medicine to treat schizophrenia and bipolar disorder. To investigate the effect of gender and polymorphisms in CYP2D6 on key pharmacokinetic parameters, two 1 mg risperidone bioequivalence studies of test/reference (Risperdal^®) formulations (n = 70 healthy Spaniards with both genders) were conducted and reanalyzed to compare the intra-individual variability of AUC_0-t and C_max [32]. Based on activity scores (AS) of the enzymes, all participants were classified into four different CYP2D6 phenotype groups as poor metabolizers (PM, AS = 0), intermediate metabolizers (IM, 0.5 < AS < 1), extensive metabolizers (EM, 1.5 < AS < 2) and ultra-rapid metabolizers (UM, AS > 2). The results reported that genetic polymorphisms considerably played a deterministic role on within-subject variability for sample size calculation over gender suggesting that volunteers participating in bioequivalence trials be genotyped. In the other standard bioequivalence trial, the two 2 mg risperidone formulations (test vs Risperdal^® tablets) were found to be bioequivalent in clinical practice with respect to the rate and extent of absorption in 24 healthy Koreans recruited according to the genotype analysis (seven homozygous for CYP2D6*1, ten for “10 and 7 heterozygous for “10) selected from around 500 subjects that were genotyped for the CYP2D6* allele after passing a clinical screening procedure [33]. To reduce the large sample size required for bioequivalence researches on highly variable drugs such as risperidone, Chen et al. [34] conducted in a different standard bioequivalence study where 20 healthy Chinese volunteers with CYP2D6 extensive metabolizers (two failed to complete) carrying the CYP2D6*10 gene were exclusively included to receive a single 4 mg oral dose of test/reference formulations (risperidone vs Risperdal^® tablets). The pharmacokinetic profiles of risperidone and its active metabolite, 9-hydroxy risperidone, after administration in study subjects with extensive metabolizers carrying CYP2D6*10 under fasting condition are observed to be interchangeable. The 90% CI for the geometric mean ratio (test/reference) of logarithm-transformed C_max, AUC_0-t and AUC_0-∞ was reported to be 91.3–118.1, 95.0–102.3 and 95.0–102.7% for risperidone, respectively, and 86.4–116.0, 83.8–109.3 and 83.6–108.8% for its active metabolite, 9-hydroxy risperidone to meet the FDA’s acceptance range. The authors projected that the cost could be saved by at least 69.5% by restricting the genotype of study subjects to reduce the sample size in the study design for developing a generic risperidone product.

The association of pharmacokinetic variability and pharmacogenomics with the bioequivalence studies of gefitinib, a tyrosine kinase inhibitor and lung cancer drug, provided by three sponsors was previously explored [35]. Two hundred and sixty male volunteers enrolled were divided into seven bioequivalence studies conducted under fasting or fed conditions to compare two 250 mg gefitinib tablets from three different generic medication manufacturers (Test, T formulation) and AstraZeneca Plc (branded, R formulation). Of these studies, two were selected and carried out for genetic analyses of CYP3A4, CYP3A5 and CYP2D6 alleles to explore the relationship between the PK outcome and CYP genetic polymorphisms. The results suggested that the pharmacokinetics of gefitinib are highly variable under fasting conditions. Three gene polymorphisms in various metabolic enzymes were investigated in two studies and three CYP2D6 rs1058164 genotypes, C/C, C/G and G/G, were found to be active and related with the drug exposures. For example, it was reported that a 39% reduction in the gefitinib AUC in CYP2D6 UMs versus EMs was observed. However, these types of IMs did not alter intra-individual variability within the selected trials and consequently the sample size for each study required to succeed bioequivalency in a standard two-period crossover study.

Tolterodine used to treat an overactive bladder is metabolized to an active 5-hydroxymethyl tolterodine by CYP2D6. Byeon et al. [36] conducted a standard bioequivalence study investigating the relationship between CYP2D6 genotypes and pharmacokinetics of tolterodine. All healthy Korean volunteers receiving 2 mg tolterodine tartrate were genotyped for CYP2D6 and divided into four different genotype groups as follows: CYP2D6*1/*2, CYP2D6*1/*10, CYP2D6*10/*10 and CYP2D6*5/*10. The coefficient of variation for each CYP2D6 genotype group and their corresponding sample sizes required to meet the power of an equivalence test were calculated. The results showed that the pharmacokinetics and within-subject variability of tolterodine were significantly associated with CYP2D6 genotypes. According to the intrasubject variation of AUC_0-t, only 26, 44 subjects in EM group and PM group, respectively, compared with around 70 subjects in the conventional study population data set were needed to meet the regulatory bioequivalence criterion. These results suggested that a drug-metabolizing enzyme genotype-based enrichment strategy can be implemented to minimize the sample size in bioequivalence studies of highly variable drug/products.

Case studies: CYP2C19

Clopidogrel, a prodrug and an antiplatelet drug to prevent heart-related diseases, is principally bioactivated to an active metabolite by CYP2C19. A standard bioequivalence study between two formulations of a 75 mg tablet of generic clopidogrel or a similar tablet of Plavix^® was conducted to investigate the effect of selected and unselected Mexican subjects homozygous for the CYP2C19*1 haplotype [37]. A total of 36 healthy volunteers of both sexes were included in the study where twenty volunteers showed to have a CYP2C19 *1/*1 homozygous genotype, two presented the *1/*17 combination of alleles, and the remaining had a *1/*2 genotype. Cedillo-Carvallo and colleagues [37] had demonstrated that exclusion of the subjects with genotypes other than the homozygous *1/*1 allele combination leaded to a slight reduction of variance and an increase in mean value for C_max, AUC_0-36h and AUC_0-∞ of both test and reference formulations. It was concluded that pharmacogenetic selection of volunteers with homozygous highly functional CYP2C19 haplotypes and removal of volunteers with less functional CYP2C19 genotypes resulted in an increase in the stringency of bioequivalence criteria and a bioinequivalence declaration for the same two formulations of clopidogrel.

Voriconazole, a triazole antifungal agent for the treatment of serious fungal infections, is extensively metabolized in human where CYP2C19 enzyme plays a key role in the N-oxidation of voriconazole. Chung et al. [38] conducted in 52 Korean volunteers, who randomly received a 200 mg intravenous infusion of either SYP-1018, a lyophilized polymeric nanoparticle, or the marketed Vfend^® formulations of voriconazole for bioequivalence trials to explore the impact of CYP2C19 polymorphism on its pharmacokinetics. A total of 51 volunteers were genotyped and 19 homozygous EMs (CYP2C19*1/*1), 19 IMs (CYP2C19*1/*2 or CYP2C19*1/*3) and 10 PMs (CYP2C19*2/*2, CYP2C19*2/*3, or CYP2C19*3/*3) were identified. It was found that the PMs yielded largest systemic drug exposure followed by the IMs, and then the EMs. CYP2C19 genotypes affected the pharmacokinetics but also the intrasubject variability of voriconazole. It was reported that for C_max, the intrasubject variability in the PMs and IMs was 44 and 22%, respectively, larger than that in the EMs and for AUC_0-t, the intrasubject variability in the PMs and IMs was 71 and 135%, respectively, larger than that in the EMs. In consequence in determining sample size, the number of subjects estimated to meet the standard bioequivalence criteria was 80 and 40% greater in the PMs and IMs for C_max, respectively and 33% and 100% greater than in the EMs for AUC_0-t, respectively.

Jiang et al. [39] had conducted a conventional bioequivalence study in 24 healthy CYP2C19 EMs (confirmed by DNA sequencing) receiving 20 mg citalopram Tablets to compare the pharmacokinetic characteristics of test (Cipramil^®) and branded (Citalopram^®) formulations to support the marketing authorization in China. CYP3A4 and CYP2C19 were known as the primary isozymes involved in the N-demethylation of citalopram as an orally administered antidepressant drug to its active metabolites, desmethylcitalopram. To avoid the additional variables participants with a CYP2C19 genotype of *1/*1 and *1/*2 along with other inclusion criteria were enrolled to take part in the study. Consequently, the test/reference formulations evaluated were pharmacokinetically equivalent within the 90% CI of the ln-transformed values of C_max, AUC_0-t, and AUC_0-∞ which met the regulatory criteria for assuming bioequivalence in the selected healthy male subjects.

Case studies: CYP3A & CYP3A5

Tacrolimus, an immunosuppressive agent, is a CYP3A substrate. It has highly variable pharmacokinetics and often fails to demonstrate bioequivalence via conventional 2 × 2 cross-over studies in 24 or less participants. To explore the potential of applying genotype-specific in a 2 × 2 cross-over bioequivalence study of 1 mg capsules of tacrolimus test/reference formulations with healthy volunteers, the effect of differing CYP3A5 genotypes (*1/*1+*1/*3, n = 16 and *3/*3, n = 13 groups) from both genders on the intrasubject variability was investigated [40]. The results showed that the intrasubject variability of AUC_0-t and C_max in the CYP3A5*3/*3 group were about 41 and 52%, respectively, greater than those in the CYP3A5*1*1+*1/*3 group and about 21 and 27%, respectively, greater than those in the total (unselected) group. Accordingly, the calculated total sample sizes projected for the bioequivalence study of tacrolimus is substantially reduced by 33% for AUC_0-t (n = 30 vs 40) and 43% for C_max (n = 28 vs 40) in the CYP3A5*1/*1+*1/*3 group compared with the total group. The results concluded that genotyping for CYP3A5 could serve as an alternative and more efficient means to support bioequivalence trials.

Case studies: CYP1A2, CYP1A1 & CYP3A4

Erlotinib, a cancer medicine, is metabolized by CYP3A4, CYP1A2, CYP1A1 and other individual CYP enzymes. Choi and co-workers [41] carried out a successful example of a two-period, two-sequence crossover bioequivalence study with healthy male Korean volunteers receiving 150 mg of erlotinib in the test/reference (Tarceva) formulations. Blood samples from 39 out of 46 male volunteers were examined for genotype analysis for eight CYP1A1, 20 CYP1A2 and 20 CYP3A4 polymorphisms. Study populations were classified based on homozygous wild-type (Wt/Wt), heterozygous type (Wt/Mt) and mutant type (Mt/Mt) for each genotype for comparison of erlotinib pharmacokinetics. Additional five genotypes, including CYP1A1*2A, CYP1A1*2B, CYP1A2*1D, CYP1A2*1M and CYP1A2*15 were measured and then classified into two subgroups by wild type and nonwild type. The results showed that genetic polymorphism in CYP1A2*1M was the metabolic enzymes alone significantly affecting erlotinib metabolism in human rather than the other genetic differences of CYP 1A1, CYP1A2 and CYP3A4. However, the data for all relevant intra-individual pharmacokinetic variabilities on AUC_0-t and C_max for each genotype were not reported to re-estimate the required sample sizes in this article.

Case studies: CYP2C9 & CYP2C8

A regular randomized, crossover study design for a 20 mg single oral dose of bioequivalent tenoxicam (an anti-inflammatory drug) test/reference (Tilcotil^®) formulations with 18 white European Spanish volunteers from both sexes was performed to explore the pharmacogenetic relevance of CYP2C9 and CYP2C8 alleles on the pharmacokinetic variability and the bioequivalence 90% classical CI [42]. Genotyping analysis for CYP2C9*2, CYP2C9*3 and CYP2C8 polymorphisms was assayed for all study subjects as follows: CYP2C9*1/1 (wild type) (n = 10), carrier of allele *2 or allele *3 (single nucleotide polymorphisms [SNPs]) (n = 8); CYP2C8*1/1 (wild type) (n = 10), carriers of alleles *3 or *4 (SNPs) (n = 8). Genotyping for CYP2C8*3 confirmed the strong linkage with CYP2C9*2 and CYP2C9*3. It was observed that high within-individual PK variation calculated in bioequivalence (AUC_0-∞) may be due to the presence of CYP2C9*3 allele and all volunteers (IC₉₀ GLOBAL, n = 18) or either selected allelic variant subgroups (IC₉₀ C9 wild type (n = 10) and IC₉₀ 2C9 SNPg, n = 8) met the preset bioequivalence range of 80–125%. Thus, for the design of bioequivalence studies of tenoxicam formulations genotype profile of the test drug should be considered.

Conclusion

This review summarizes all the pharmacogenetic–pharmacokinetic methods available from the literature for bioequivalence assessment of highly variable drugs/products. The difficulty of fulfilling bioequivalence in HV drug/products is in general the rather large of sample size using a standard two-period crossover study design. Herein, the authors presented adopting pharmacogenetic approaches to minimize the within-subject variability executed in a standard crossover bioequivalence study. It was well-documented that intrasubject pharmacokinetic variability can be impacted through gene expression regulation at both an individual and population level. However, application of pharmacogenetics in bioequivalence studies has been considered relatively little up to now. This article aims to begin with an overview on the ways as a research priority in which use of pharmacogenetic screen is performed to reduce sample size while maintaining standard study design and bioequivalence acceptance criteria. We believe that it would be beneficial to further widen pharmacogenetic–pharmacokinetic applications in bioequivalence evaluation areas considering potential savings to patients in big fortunes after approval of HV generic products.

Future perspective

The initial attempt of including pharmacokinetics-related pharmacogenetic screen in bioequivalence studies for generic drug development was to explore whether clinical genetic variation in a subset of the population may respond differently to the two formulations of highly variable generic products to such an extent that this would lead to a decline in the recruitment of sample sizes required to meet adequate statistical power. Future implement of quality by design and appropriate in vitro dissolution testing together with physiologically-based pharmacokinetic modeling systems to reduce subject variance could certainly result in cost saving by improving the efficiency of bioequivalence trials and reducing the clinical studies size.

The impact of transporter polymorphism and non-CYP Phase II-metabolizing enzymes [60–63] on altering drug pharmacokinetics is evident to be of less significance compared with CYP polymorphic enzymes but it should not be totally ignored as an extension of our knowledge in drug transporters being underway. Future advancement of new technologies in genome sequencing and completed absorption, distribution, metabolism and elimination gene analyses might play a significant role in investigating pharmacogenetic–pharmacokinetic–bioequivalence studies for highly variable generics.

In the case of regulatory discrepancy, for instance, the EMA and the FDA reflect different claims toward selecting subjects to execute bioequivalence trials [64]. The guidance from the EMA suggested: ‘the subject population for bioequivalence studies should be selected with the aim to minimize variability’ while the guidance from the FDA indicated: “in general, … In vivo bioequivalence study subjects should be representative of the general population, taking into account age, sex and race…” . There is a lack of global consensus with regard to the inclusion or exclusion criteria of poor versus EMs as an effective pharmacogenetic approach to further minimize pharmacokinetic intra-individual variability in support of official bioequivalence designs.

Executive summary

Background & rationale for efficient pharmacogenetic-pharmacokinetic-bioequivalence studies for highly variable drugs

• In a similar way to employ younger healthy volunteers, pharmacogenetic screen of volunteers might minimize the subject variability of pharmacokinetic outcomes in human and subsequently increase stringency of bioequivalence studies.

• Clinical pharmacogenetic–pharmacokinetic–bioequivalence trials may help efficiently recruit smaller subject numbers executing a standard 2 × 2 design without being outweighed by an increased number of periods to avoid carryover (from preceding formulation) effects as compared with the replicate designs recommended by regulatory agencies for highly variable drugs/products.

Existing bioequivalence studies

• Polymorphisms in genotypes influence the pharmacokinetics/bioequivalence results of highly variable test/reference drugs.

• There is a tendency to employ subpopulations via pharmacogenetically homogeneous selection of study subjects to allow better clinical pharmacokinetic profiles for bioequivalence outcomes.

• Key CYP enzyme activities are associated with intra-individual variation measures and the degree of sample size reduction for the bioequivalence trials of HV test/reference formulations.

Financial & competing interests disclosure

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.