Bioequivalence of two rabeprazole formulations and pharmacokinetic variability by CYP2C19 genotype and sex in healthy Korean participants: A four‐period replicated crossover study

Abstract

Rabeprazole is a highly variable drug for which demonstration of bioequivalence is often challenging, and CYP2C19 genotype and sex have been suggested as potential contributors to pharmacokinetic variability. We evaluated the pharmacokinetics, bioequivalence (BE), and tolerability of a generic rabeprazole formulation compared with the reference product in healthy Korean participants using an open‐label, randomized, single‐dose, two‐sequence, four‐period replicated crossover design with a scaled average BE (SABE) approach. Exploratory subgroup analyses according to CYP2C19 genotype and sex were also performed. Fifty‐two participants received single 20 mg doses under fasting conditions, and 46 completed the trial. BE was evaluated using SABE and conventional average BE (ABE), with subgroup analyses by genotype and sex. Geometric mean ratios (GMRs) of the test to reference formulations were within the BE acceptance range under ABE and SABE approaches. The reference formulation showed high intra‐CV in maximum plasma concentration (C _max; 67.9%), consistent with the classification of rabeprazole as a highly variable drug. These results demonstrated bioequivalence between the test and reference formulations under both conventional average bioequivalence and scaled bioequivalence approaches.

Rabeprazole, a proton pump inhibitor (PPI), was discovered and developed by the Japanese pharmaceutical company Eisai Co., Ltd. It was first launched in Japan in 1997 under the brand name Pariet, followed by releases in Europe in 1998 and in the United States in 1999, where it was marketed as Aciphex. ¹ Today, rabeprazole is approved in more than 100 countries for the treatment of gastroesophageal reflux disease, duodenal ulcers, and other acid‐related disorders. ²

The patent for rabeprazole, titled “Pyridine derivatives having anti‐ulcerative activity,” expired on May 8, 2013, and generic formulations have been available since. Regulatory approval for generics requires demonstration of bioequivalence (BE), defined as the absence of a significant difference in the rate and extent of absorption of the active ingredient between test and reference products.

Rabeprazole is a highly variable drug with an intrasubject coefficient of variation (intra‐CV) in its maximum plasma concentration (C _max) exceeding 30%. According to the U.S. Food and Drug Administration Draft Guidance on Rabeprazole Sodium, in addition to the average BE (ABE) approach, a scaled bioequivalence approach may also be applied for rabeprazole. ³ A preliminary analysis of data from the Korea Ministry of Food and Drug Safety (KMFDS) drug safety information website reported that a BE study of rabeprazole 20 mg (Test product: Pabizol 20 mg; Reference product: Pariet 20 mg), conducted using a 2 × 4 crossover design, showed an intra‐CV for C _max of 56%. ⁴ Furthermore, an analysis of BE studies conducted by Invites Biocore Co., Ltd., a clinical research organization specializing in clinical bioanalysis, revealed a success rate of only 30% (6 out of 18 studies, internal data), underscoring the difficulties associated with demonstrating BE for this highly variable drug.

Rabeprazole is highly unstable when exposed to moisture, leading to significant degradation, particularly in acidic environments. This intrinsic instability of rabeprazole presents a major challenge in research and manufacturing. Rabeprazole is formulated exclusively as an enteric‐coated formulation to overcome gastric acid degradation. ⁵ This physicochemical property of rabeprazole also complicates in vitro dissolution studies that are essential for preliminary evaluation before BE testing. Specifically, reliable dissolution profiling in pH 1.2 and 4.5 buffer media cannot be achieved due to the enteric coating. This limitation hinders accurate assessment of the dissolution behavior, further complicating formulation optimization and quality control. Ultimately, rabeprazole is a compound that is exceptionally difficult to handle in research and development; thus, establishing BE reflects advanced formulation technology and robust manufacturing capability.

Allelic variants of the CYP2C19 gene, such as *2 (c.681G>A, rs4244285), *3 (c.636G>A, rs4986893), and *17 (c.−806C>T, rs12248560), have been well characterized in pharmacogenetic research, with *1 considered the wild‐type allele. Based on these genotypes, the Clinical Pharmacogenetics Implementation Consortium categorizes individuals into metabolizer phenotypes as follows: *17/*17, ultrarapid metabolizers (UMs); *1/*17, rapid metabolizers (RMs); *1/*1, normal metabolizers (NMs); *1/*2 and *1/*3, intermediate metabolizers (IMs); and *2/*2, *2/*3, and *3/*3, poor metabolizers (PMs). ⁶

Similar to other PPIs, the metabolism of rabeprazole is influenced by the CYP2C19 enzyme; however, rabeprazole has the least dependence on CYP2C19 among PPIs. ⁷ , ⁸ , ⁹ , ¹⁰ Unlike other PPIs, the primary metabolic pathway of rabeprazole involves the nonenzymatic reduction to form thioether compounds. ⁷ Only a small portion of rabeprazole metabolism is influenced by CYP2C19 and CYP3A4, producing demethylated rabeprazole and rabeprazole sulfone as minor metabolites. ⁷ Consequently, the relative impact of the CYP2C19 pathway on the metabolism of PPIs has been reported in the following order: omeprazole = esomeprazole > pantoprazole > lansoprazole > rabeprazole. ⁸ The metabolism of rabeprazole is less affected by genetic polymorphisms of this enzyme due to its reduced reliance on CYP2C19. ⁷ , ⁸ , ⁹ , ¹⁰ The impact of CYP2C19 polymorphisms on rabeprazole pharmacokinetics (PKs) has been investigated in Japanese and Chinese open‐label, randomized, crossover studies comparing rabeprazole with omeprazole and lansoprazole, which consistently demonstrated that plasma exposure increased with reduced CYP2C19 enzymatic activity (PMs > IMs > NMs), but the extent of this effect was smallest for rabeprazole. ¹¹ , ¹² Collectively, these results suggest that rabeprazole is least affected by CYP2C19 polymorphisms, indicating potential benefits in variant‐prevalent populations.

In a Korean cohort (N = 432), phenotype frequencies of NMs 40%, IMs 45%, PMs 13%, and UMs 2%. ¹³ According to Zhang et al., the frequency of these phenotypes varies considerably across ethnic groups. ¹⁴ The PM phenotype occurs in approximately 1%–5% of Caucasians, 3%–4% of Africans, and 10%–23% of Asians, while the IM phenotype is reported at 22%–30% in Caucasians, 23%–49% in Africans, and 35%–55% in Asians. ¹⁴ The prevalence of IM and PM phenotypes is higher in Koreans than in Caucasians (45% vs. 22%–30% and 13% vs. 1%–5%, respectively).

Although the study by Chen et al. assessed BE using a two‐sequence, two‐period (2 × 2) crossover design in Chinese participants, recent studies adopting a two‐sequence, four‐period (2 × 4) fully replicated crossover design for this highly variable drug remain limited to publicly accessible databases. ¹⁵ This study underscores the lack of comprehensive PK data for rabeprazole obtained using advanced study designs. Accordingly, the primary objective of this study was to compare the PK and tolerability profiles of a generic formulation of rabeprazole with those of Pariet 20 mg, the reference product, using the SABE method in a 2 × 4 crossover study. Despite numerous studies on rabeprazole, to date, no study has characterized the PK variations attributable to the CYP2C19 genotype in the Korean population. Therefore, as a secondary objective, this study included a subgroup analysis based on CYP2C19 genotype and sex to explore the PK characteristics in Korean participants.

METHODS

Study design and ethics

This study was an open‐label, randomized, single‐dose, two‐sequence, four‐period replicated crossover regulatory BE study conducted in healthy participants under fasting conditions. The protocol and informed consent forms were reviewed and approved by the Institutional Review Board of H+ Yangji Hospital, Seoul, Republic of Korea (IRB Nos. HYJ 2021‐10‐030‐012 and HYJ 2023‐06‐031‐001) and the Ministry of Food and Drug Safety. The study was registered in the WHO‐recognized Clinical Research Information Service (CRIS; KCT0011220). All study procedures were conducted in accordance with the approved protocol, the Declaration of Helsinki, International Council for Harmonization Good Clinical Practice (ICH‐GCP) guidelines, and applicable institutional standard operating procedures.

The sample size was calculated for 90% power at a 5% significance level, assuming an intra‐CV of 56% for C _max and a test/reference (T/R) ratio of 1.1, yielding 36 participants. To account for a 30% dropout rate, 52 participants were enrolled and randomized to either sequence A (RTRT) or B (TRTR). Each participant received a single 20 mg dose of rabeprazole under fasting conditions, with a 1‐week washout between periods. The 20 mg dose was selected because it corresponds to a clinically relevant marketed strength of rabeprazole and is commonly used in bioequivalence studies and clinical practice.

Participants

Healthy Korean adults aged ≥19 years with a body mass index of 18–30 kg/m² and minimum body weights of 50 kg for men and 45 kg for women were enrolled after meeting predefined inclusion and exclusion criteria. All participants were in good health, free of clinically significant congenital or chronic diseases, and had no abnormal findings on standard medical evaluations. Key exclusion criteria included recent participation in other clinical trials, history of gastrointestinal surgery, use of enzyme‐inducing or enzyme‐inhibiting drugs, regular alcohol consumption, known hypersensitivity to rabeprazole or related compounds, and concurrent use of contraindicated medications. Final eligibility was confirmed by the investigator based on clinical judgment.

Drug formulations (in vitro dissolution)

Because rabeprazole is an acid‐labile enteric‐coated formulation, dissolution testing was performed under intestinal‐like conditions (pH 6.0) to confirm comparable release characteristics between formulations. Both formulations showed comparable dissolution profiles (Figure S1). Detailed experimental conditions are provided in Supporting Information.

Drug administration and blood sampling

Drug administration and blood sampling were conducted at the Clinical Research Center of H+ Yangji Hospital in Seoul. Before study initiation, participants received a comprehensive explanation of the study's objectives, procedures, and potential risks from the principal investigator or a delegated investigator. Written informed consent was obtained, signed, and dated by both parties. Participants were hospitalized the day before dosing, fasted for at least 10 h, and received the investigational product with 150 mL of water at 8:00 AM after baseline assessments. Water intake was restricted from 1 h before to 1 h after dosing. Clinical assessments, including physical examinations, vital signs, and laboratory tests, were performed according to the manufacturer's protocol. Adverse events (AEs) were continuously monitored and documented.

Blood samples were collected over a 12‐h period to meet the KMFDS Standards for Pharmaceutical Equivalence Testing, which require covering at least three times the elimination half‐life or 80% of the area under the concentration–time curve (AUC) from time zero to the last quantifiable concentration (AUC_last)/AUC from dosing to infinity (AUC_inf). ¹⁶ A total of 17 samples (5 mL each) were collected in EDTA‐K₂ vacutainers at predefined intervals: 0, 1, 1.5, 2, 2.5, 3, 3.5, 4, 4.5, 5, 5.5, 6, 6.5, 7, 8, 10, and 12 h. Samples were immediately centrifuged at 3000 rpm for 10 min at 4 °C. An aliquot of plasma (0.8 mL) was transferred into two labeled polypropylene tubes containing 25 µL of 1% diethylamine, vortex‐mixed, and stored at −70 °C until PK analysis.

Determination of rabeprazole plasma concentrations

Quantitative analysis of rabeprazole plasma concentrations was conducted by Invites BioCore Co., Ltd. using high‐performance liquid chromatography (Shimadzu LC‐40, Japan) coupled with tandem mass spectrometry (TQ5500+, SCIEX, USA). In total, 3239 samples were processed, with rabeprazole‐d4 used as the internal standard. Plasma samples were prepared via protein precipitation and analyzed using a C18 column (2.0 mm × 75 mm, 3 µm) with a mobile phase of 60% 10 mm ammonium acetate and 40% acetonitrile containing 0.01% ammonium hydroxide at a flow rate of 0.3 mL/min. The assay had a lower limit of quantification of 1 ng/mL and a calibration range of 1–1000 ng/mL. Quality control samples at 3300 ng/mL and 800 ng/mL demonstrated accuracy between 94.7% and 106.4%, with coefficients of variation below 5%, confirming high analytical precision and reproducibility.

PK analysis

PK parameters were calculated using a noncompartmental method with Phoenix WinNonlin version 8.3 (Certara, St. Louis, MO, USA). C _max and T _max were derived directly from the individual plasma concentration–time (C–T) profiles of rabeprazole. AUC_last was calculated using the linear trapezoidal method. AUC_inf was calculated as the sum of AUC_last and C _last/λ_z , where C _last represents the last measurable concentration and λ_z is the terminal elimination rate constant estimated from the terminal log‐linear portion of the C–T curve. The terminal elimination half‐life (t _½) was determined as ln(2)/λ_{z .}

Genotyping

CYP2C19 genotyping was performed in a subset of participants (N = 24) who provided additional consent for genetic analysis. Plasma samples collected at 0 h during the BE study were used, eliminating the need for additional specimen collection. DNA was extracted and amplified using a ProFlex PCR System (Applied Biosystems), followed by sequence analysis using a 3500xLDX Genetic Analyzer (Applied Biosystems). Mutations c.681G>A (*2) and c.636G>A (*3) in CYP2C19 were identified using Sequencher software. Based on the genotype results, the participants were categorized as NMs (*1/*1), IMs (*1/*2 or *1/*3), or PMs (*2/*2, *2/*3, or *3/*3).

Statistical analysis

PK statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA). Log‐transformed PK parameters (C _max, AUC_last, and AUC_inf) were analyzed using an analysis of variance (ANOVA) model including sequence, period, and treatment as fixed effects and subject nested within sequence as a random effect. BE was evaluated based on the log‐transformed C _max, AUC_last, and AUC_inf values using a 90% confidence interval (CI) for the geometric mean ratio (GMR). BE was considered if the 90% CI fell within the conventional range of 0.80–1.25. When the intra‐CV for C _max of the reference formulation exceeded 30%, the SABE approach was applied in accordance with the KMFDS guideline for highly variable drugs, which is consistent with the European Medicines Agency (EMA) framework. In this approach, the acceptance range for C _max was expanded to 0.6984–1.4319 (intra‐CV ≥50%) in accordance with the European Medicines Agency ¹⁷ and KMFDS guidelines. ¹⁶ The scaling was calculated as [L, U] = exp(±k·SWR), where k = 0.760 and SWR is the within‐subject standard deviation of log‐transformed C _max values. AUC_last and AUC_inf were required to remain within the standard range of 0.80–1.25.

The 95% CI for adverse drug reaction incidence was calculated, and events were classified using the Preferred Term within each System Organ Class. The severity and causality were assessed. Descriptive statistics were used to summarize the baseline demographics (age, sex, height, weight, and body mass index), vital signs, and laboratory parameters.

RESULTS

Participants and genotyping

A total of 95 healthy adult volunteers provided informed consent and were screened for eligibility. Among them, 15 did not meet the inclusion and exclusion criteria, 17 withdrew their consent before participation, and 11 were enrolled as reserve participants, resulting in a total of 43 exclusions. Ultimately, 52 participants received at least one dose of the investigational drug; 46 completed the study, and six discontinued after dosing. The flow of participants through the study is shown in Figure 1. Baseline demographic characteristics by treatment sequence are summarized in Table 1. Among the 46 participants who completed the study, 31 were male and 15 were female. Twenty‐four participants (52%) consented to CYP2C19 genotyping, which identified 15 NMs (*1/*1) and nine IMs, including *1/*2 (n = 4) and *1/*3 (n = 5), as shown in Table S1. A statistically significant difference in weight was observed between the sequence groups; however, no significant differences were observed in other demographic variables.

Figure 1.

Flow of participants through the study.

Table 1.

Baseline demographic characteristics of study participants by the treatment sequence group.

	Sequence		Total
	R‐T‐R‐T	T‐R‐T‐R
	(N = 26)	(N = 26)	(N = 52)	p‐value
Sex
Male	21 (80.77%)	15 (57.69%)	36 (69.23%)	0.0714 a
Female	5 (19.23%)	11 (42.31%)	16 (30.77%)
Phenotype
NM	7 (26.92%)	8 (30.77%)	15 (28.85%)
IM	4 (15.38%)	5 (19.23%)	9 (17.31%)
Unknown	15 (57.69%)	13 (50.00%)	28 (53.85%)
Age (years)	26.27 ± 8.45	26.77 ± 6.62	26.52 ± 7.52	0.5437 b
Height (cm)	171.52 ± 7.55	167.73 ± 8.67	169.63 ± 8.27	0.0989 c
Weight (kg)	69.39 ± 11.23	63.64 ± 9.3	66.52 ± 10.61	0.0499 * c
BMI (kg/m2)	23.47 ± 2.61	22.57 ± 2.68	23.02 ± 2.66	0.2276 c

T, test product: Daewoong rabeprazole tab. 20 mg; R, reference product: Pariet 20 mg.

BMI, body‐mass index; NM, normal metabolizer; IM, intermediate metabolizer.

^a Chi‐square test,

^b Wilcoxon's rank sum test,

^c Independent t‐test,

^* p < 0.05, statistically significant.

PK parameters and BE

Among the 52 participants initially enrolled, 46 completed the study and were included in the PK analysis. Under fasting conditions, the PK profiles of the test and reference rabeprazole formulations were comparable (Table 2 and Table S2). The mean values for C _max, AUC_last, and AUC_inf were similar between the two formulations, with median T _max ranging from 4.0–4.5 h. Although slight differences in t _½ were observed, the variability was consistent with the known pharmacokinetic variability of rabeprazole. Overall, the test and reference formulations demonstrated comparable absorption and elimination characteristics, supporting bioequivalence between the two products. The mean (± standard deviation) plasma concentration–time profiles are shown in Figure 2.

Table 2.

Summary of rabeprazole pharmacokinetic parameters following the administration of 2 single doses of 2 formulations under fasted conditions.

Parameters	Test product			Reference product
	First administration (N = 46)	Second administration (N = 46)	Total (N = 46)	First administration (N = 46)	Second administration (N = 46)	Total (N = 46)
C max (ng/mL)	849.68 ± 422.96	810.82 ± 450.27	830.25 ± 434.86	800.39 ± 414.81	849.02 ± 419.65	824.71 ± 415.66
AUClast (h·ng/mL)	1552.08 ± 820.17	1542.79 ± 785.11	1547.49 ± 798.57	1487.01 ± 758.42	1597.21 ± 801.90	1542.11 ± 778.14
AUCinf (h·ng/mL)	1644.47 ± 864.03	1667.74 ± 797.41	1656.23 ± 826.35	1857.84 ± 1419.82	1669.79 ± 842.79	1661.74 ± 790.41
T max (h)	4.0 [2.0−12.0]	3.8 [0.0−6.5]	4.0 [0.0−12.0]	4.50 [2.0−10.0]	4.0 [2.5−8.0]	4.5 [2.0−10.0]
t ½ (h)	2.04 ± 0.76	2.91 ± 4.05	2.48 ± 2.95	1.78 ± 0.45	1.96 ± 0.67	1.87 ± 0.58

Values are presented as arithmetic mean ± standard deviation, except for T _max, which is presented as median (range); test product: Daewoong rabeprazole tab. 20 mg; reference product: Pariet 20 mg.

AUC_inf, area under the plasma drug concentration‐time curve from time 0 to infinity; AUC_last, area under the concentration‐time curve from time zero to time of last measurable concentration; C _max, maximum plasma concentration; T _max, the time to peak plasma concentration; t _½, elimination half‐life.

Figure 2.

Plasma concentration‐time curves of rabeprazole. Reference and test drug in each period: (A) linear and (B) log scale. The error bars represent the standard deviations. R, reference rabeprazole; T, test rabeprazole.

PK parameters were analyzed using an ANOVA model to assess significant differences among sequences, periods, and treatment effects at a significance level of 0.05. No significant effects of sequence, period, or treatment were observed (all P > 0.05). The GMRs of the test to reference formulations were 1.0124 (90% CI: 0.8768–1.1690) for C _max, 1.0116 (90% CI: 0.9297–1.1008) for AUC_last, and 0.9859 (90% CI: 0.9123–1.0655) for AUC_inf. All 90% CIs were within the conventional acceptance range of 0.80–1.25.

Because the intra‐CV for C _max of the reference formulation was 67.9%, the SABE approach was also applied. The C _max values fell within the widened SABE limits of 0.6984–1.4319, as permitted by regulatory standards. These results demonstrated that the test and reference formulations were bioequivalent under fasting conditions based on both the conventional ABE and SABE criteria. The GMRs for C _max, AUC_last, and AUC_inf were close to unity, indicating comparable systemic exposure between the test and reference formulations (Table 3).

Table 3.

Summary of pharmacokinetic parameters and bioequivalence evaluation by overall group and subgroups.

			Geometric means a
Group	N	Parameter	Reference	Test	GMR b (90% CI)	Intra CV % c	Equivalent criteria d	Evaluation
Overall group
All	46	C max (ng/mL)	689.9	695.8	1.0124 (0.8768−1.1690)	67.9	0.6984−1.4319	Bioequivalent
		AUClast (h·ng/mL)	1343.4	1350.3	1.0116 (0.9297−1.1008)	37.5	0.8000−1.2500	Bioequivalent
		AUCinf (h·ng/mL)	1530.8	1463.7	0.9859 (0.9123−1.0655)	32.4	0.8000−1.2500	Bioequivalent
Subgroup
NM	15	C max (ng/mL)	767.5	585.4	0.7722 (0.5786−1.0307)	40.2	0.7450−1.3424	Did not meet BE criteria
		AUClast (h·ng/mL)	1371.3	1167.4	0.8755 (0.7263−1.0552)	24.7	0.8000−1.2500	Did not meet BE criteria
		AUCinf (h·ng/mL)	1403.1	1314.5	0.9690 (0.8178−1.1482)	23.5	0.8000−1.2500	Bioequivalent
IM	9	C max (ng/mL)	634.6	652.9	1.0325 (0.6534−1.6316)	100.4	0.6984−1.4319	Did not meet BE criteria
		AUClast (h·ng/mL)	1285.3	1368.0	1.0614 (0.8275−1.3614)	51.8	0.8000−1.2500	Did not meet BE criteria
		AUCinf (h·ng/mL)	1831.3	1473.2	0.7976 (0.5951−1.0689)	57.7	0.8000−1.2500	Did not meet BE criteria
Male	31	C max (ng/mL)	641.8	676.2	1.0748 (0.8977−1.2869)	70.8	0.6984−1.4319	Bioequivalent
		AUClast (h·ng/mL)	1331.1	1402.0	1.0757 (0.9736−1.1884)	39.2	0.8000−1.2500	Bioequivalent
		AUCinf (h·ng/mL)	1468.5	1517.2	1.0536 (0.9728−1.1410)	22.0	0.8000−1.2500	Bioequivalent
Female	15	C max (ng/mL)	801.2	737.4	0.8636 (0.6461−1.1543)	54.9	0.6984−1.4319	Did not meet BE criteria
		AUClast (h·ng/mL)	1369.4	1251.0	0.8612 (0.7193−1.0311)	32.3	0.8000−1.2500	Did not meet BE criteria
		AUCinf (h·ng/mL)	1670.9	1353.6	0.8253 (0.6898−0.9874)	48.2	0.8000−1.2500	Did not meet BE criteria

In the 2 × 4 crossover design, each participant received two administrations of both formulations; therefore, the total number of observations for statistical analysis was doubled (46 × 2 = 92).

AUC_inf, area under the plasma drug concentration‐time curve from time 0 to infinity; AUC_last, area under the concentration‐time curve from time zero to time of last measurable concentration; CI, confidence interval; C _max, maximum plasma concentration; NM, normal metabolizer; GMR, geometric mean ratio; IM, intermediate metabolizer; Intra CV, Intra‐individual coefficient of variation.

^a Geometric least squares means (GLSMs);

^b Test to reference;

^c Intra‐subject CV on reference;

^d Scaled according to the scaled average bioequivalence (SABE) approach; performed using SAS software version 9.4.

Subgroup analysis

The PK parameters and bioequivalence of rabeprazole were evaluated in subgroups stratified by CYP2C19 genotype (NM and IM) and sex (male and female).

CYP2C19 genotype subgroup analysis (NM: 15, IM: 9)

In the genotype‐based subgroup analysis, no significant differences were observed in C _max, AUC_last, T _max, or t _½ between the NM and IM groups for the reference product. However, AUC_inf in the IM group was 1.40‐fold higher than in the NM group (2259.60 ± 2055.83 vs. 1613.31 ± 913.00 h·ng/mL), with considerable variability among IMs (Table S2). The mean (± standard deviation) plasma C–T curves are shown in Figure S2A,B, where T _max appears slightly delayed in the IM group.

The NM group showed the lowest intra‐CVs (C _max, 40.2%; AUC_last, 24.7%; AUC_inf, 23.5%). In the NM group, the CIs for AUC_inf (GMR: 0.9690; 90% CI: 0.8178–1.1482) remained within the conventional bioequivalence acceptance range, whereas the CIs for C _max and AUC_last extended beyond the conventional bioequivalence acceptance range (GMR: 0.7722; 90% CI: 0.5786–1.0307 and GMR: 0.8755; 90% CI: 0.7263–1.0552, respectively). In the IM group, the CIs for PK parameters were wider and generally exceeded the conventional bioequivalence acceptance range, with intra‐CVs of 100.4%, 51.8%, and 57.7% for C _max, AUC_last, and AUC_inf, respectively. Because these exploratory subgroup analyses were conducted with limited sample sizes, the corresponding CIs should be interpreted cautiously (Table 3).

Sex‐based subgroup analysis (male: 31, female: 15)

In the sex‐based analysis, AUC_last values were comparable between male and female participants. However, female participants exhibited higher C _max and AUC_inf values (1.24‐fold and 1.23‐fold, respectively), suggesting increased systemic exposure (Table S2). Mean body weight was lower in female participants than in male participants (56.0 ± 7.2 vs. 71.3 ± 7.9 kg; Table S3), which may have partially contributed to the observed differences in systemic exposure. Figure S2C,D show consistently higher plasma concentrations in female participants. Previous studies have reported that sex does not significantly influence CYP2C19 polymorphisms. ¹⁸ , ¹⁹ In the present study, IMs were proportionally more frequent among male participants than female participants (Table S3). Therefore, the higher systemic exposure observed in female participants is unlikely to be explained solely by differences in CYP2C19 phenotype distribution and may reflect other sex‐related physiological factors.

In male participants, the CIs for C _max, AUC_last, and AUC_inf remained within the conventional bioequivalence acceptance range (GMR: 1.0748; 90% CI: 0.8977–1.2869; GMR: 1.0757; 90% CI: 0.9736–1.1884; and GMR: 1.0536; 90% CI: 0.9728–1.1410, respectively). In female participants, the CIs for PK parameters were wider and generally extended beyond the conventional bioequivalence acceptance range (GMR: 0.8636; 90% CI: 0.6461–1.1543 for C _max; GMR: 0.8612; 90% CI: 0.7193–1.0311 for AUC_last; and GMR: 0.8253; 90% CI: 0.6898–0.9874 for AUC_inf). Similar to the genotype subgroup analysis, these exploratory subgroup analyses were conducted with limited sample sizes; therefore, the corresponding CIs should be interpreted cautiously (Table 3).

Safety

A total of 49 and 50 participants received the test (T) and reference (R) products, respectively. Each participant received two administrations under the 2 × 4 crossover design, resulting in 95 and 96 administrations evaluated for safety in the T and R groups, respectively. No serious AEs occurred. In the T group, one participant (1.05%) experienced two adverse drug reactions, nausea and headache, both related to the study drug. In the R group, two AEs were reported by different participants (2.08%), including one case of skin abrasion and one positive SARS‐CoV‐2 test result, both considered unrelated to the investigational product. All AEs resolved, and all patients recovered. Overall, both formulations were well tolerated with no significant safety concerns (Table S4).

DISCUSSION

The primary objective of this study was to evaluate the PK and safety profiles of a generic rabeprazole formulation compared with the reference product, Pariet 20 mg. In a randomized, open‐label, four‐period crossover study involving healthy Korean participants under fasting conditions, bioequivalence was demonstrated using both the ABE and SABE criteria. The GMRs for C _max, AUC_last, and AUC_inf were within conventional acceptance limits, and the high intra‐CV C _max (67.9%) justified the use of the SABE method. Both formulations were well tolerated with no serious or clinically significant AEs reported, confirming their comparable safety and PK profiles.

Rabeprazole exhibited high within‐subject variability in this study, with an intra‐CV of 67.9% for C _max of the reference formulation, which was more than twice the regulatory threshold of 30% for applying the SABE approach. High intra‐CV observed in replicated crossover studies may arise from multiple physiological and formulation‐related factors, including variability in gastric emptying, gastrointestinal pH, enteric coating disintegration, and absorption timing, particularly for acid‐labile enteric‐coated formulations such as rabeprazole. Previous studies of other CYP‐metabolized drugs have also reported differences in systemic exposure according to CYP2C19 genotype. ²⁰ , ²¹ For example, Chung et al. (2015) reported that in Korean participants, both systemic exposure and intra‐CV in C _max and AUC_inf of voriconazole were highest in PMs, followed by IMs and NMs, correlating with the CYP2C19 genotype. ²⁰ A similar trend was observed for tacrolimus, where intra‐CV was significantly greater in CYP3A5 PMs than in NM and IM groups. ²¹ However, these observations were derived from conventional 2×2 crossover studies and do not directly explain the within‐subject variability measured as intra‐CV in a replicated crossover design. Accordingly, the subgroup findings in this study should be interpreted cautiously and considered hypothesis‐generating observations that warrant further investigation in prospectively designed studies.

In this study, no significant differences were observed between NMs (n = 15) and IMs (n = 9) in terms of C _max, AUC_last, T _max, or t _½, with relative values close to unity (C _max 1:0.94, AUC_last 1:0.97, T _max 1:1.1, t _½ 1:1). However, the AUC_inf of the IMs was approximately 1.4‐fold higher than that of the NMs (Table S2). These findings are consistent with prior Japanese and Chinese crossover studies evaluating rabeprazole, omeprazole, and lansoprazole. ¹¹ , ¹² Sakai et al. reported that in Japanese participants (N = 6 per group; rabeprazole 20 mg), relative AUC_inf values across NMs, IMs, and PMs were 1:1.1:1.2, with corresponding C _max and t _½ ratios of 1:1.1:0.9 and 1:1.2:1.4, respectively. ¹¹ Similarly, Qiao et al. found in Chinese participants (N = 6 per group; rabeprazole 40 mg) that relative AUC_inf values were 1:1.6:3.7, C _max and t _½ ratios of 1:1.2:2.0 and 1:1.4:1.9, respectively. ¹² In both studies, NM and IM groups showed only modest differences in rabeprazole exposure, with the largest increases observed in PMs. Our findings mirror this pattern, demonstrating that although IMs exhibited a slightly higher AUC_inf than NMs, the effect was relatively small. Collectively, these results reaffirm that rabeprazole is less susceptible to CYP2C19 polymorphisms than omeprazole or lansoprazole, supporting its PK stability in populations with a high prevalence of CYP2C19 variants, such as East Asians.

In this exploratory analysis, the intra‐CV values in the IM group were more than twice those observed in the NM group across all PK parameters. Specifically, the intra‐CV for C _max was 100.4% in IMs compared with 40.2% in NMs, whereas AUC_last and AUC_inf showed intra‐CVs of 51.8% and 57.7% in IMs versus 24.7% and 23.5% in NMs, respectively (Table 3). Additionally, Figure 3A,B illustrate greater interindividual variability in IMs, as shown by the broader interquartile ranges in the boxplots. These observations may suggest potential differences in rabeprazole exposure according to CYP2C19 phenotype; however, the exploratory nature of the subgroup analysis and limited sample size preclude definitive conclusions.

Figure 3.

Boxplot comparison of rabeprazole C _max and AUClast according to CYP2C19 genotype and sex. Panels show (A) C _max by genotype [normal metabolizers (NM) vs. intermediate metabolizers (IM)], (B) AUClast by genotype (NM vs. IM), (C) C _max by sex (male vs. female), and (D) AUClast by sex (male vs. female). Data are presented as median and interquartile range (IQR). Boxes represent the 25th to 75th percentiles, the line within each box indicates the median, whiskers extend to 1.5 × IQR, and dots represent individual outliers.

Furthermore, because a pre‐specified plan for CYP2C19 genotype evaluation was not included in the original study design, genotyping was performed in only 24 participants, representing approximately half of the total study population (N = 46). Notably, no PMs were identified. These limitations underscore the need for further studies to assess the effects of CYP2C19 polymorphisms on PK variability comprehensively. Although many existing studies have primarily focused on PMs, NMs, and IMs constitute approximately 85% of the Korean population, with a nearly 1:1 distribution (NMs: 40%, IMs: 45%). Therefore, the findings from the IM group may provide valuable insights for designing future PK or BE studies of rabeprazole.

Subgroup analysis was also conducted to investigate the influence of sex on PK variability. AUC_last values were comparable between male and female participants. Female participants exhibited higher C _max and AUC_inf values, with 1.24‐fold and 1.23‐fold increases, respectively, suggesting increased systemic exposure. A similar observation was reported by Jeong et al. in a study of Korean participants (male, n = 24; female, n = 21) who received 10 mg rabeprazole. ²² In that study, C _max was 1.20 times higher in female participants, although AUC_last and AUC_inf varied by less than 15% between sexes and were not statistically significant. The difference in AUC_inf findings between the two studies may partly reflect differences in dose level, which could influence the magnitude of sex‐related differences in systemic exposure.

Intra‐CV was further evaluated by sex. In female participants, the intra‐CVs for C _max and AUC_last were 54.9% and 32.3%, respectively, whereas in male participants, they were 70.8% and 39.2%, respectively, indicating high variability in both groups. The intra‐CV for AUC_inf was more than twofold higher in female participants (48.2%) than in male participants (22.0%), indicating greater variability in systemic exposure among females. Figure 3C,D illustrate greater interindividual variability in the female group, as evidenced by the broader interquartile ranges in the boxplots. These findings may indicate potential sex‐related differences in systemic exposure to rabeprazole.

Although exploratory subgroup analyses suggested differences in systemic exposure according to CYP2C19 genotype and sex, these between‐subject factors cannot directly explain the intra‐CV observed in this replicated crossover study. Intra‐CV in rabeprazole pharmacokinetics may also reflect multifactorial physiological and formulation‐related influences, including variability in gastric emptying, enteric‐coating disintegration, gastric pH transition, and intestinal transit. In the present study, dissolution profiles of the test and reference formulations were comparable under the selected dissolution conditions (Figure S1), and dosing conditions, including water intake, were standardized across periods. In addition, period‐specific plasma concentration–time profiles did not demonstrate an obvious systematic trend across administrations (Figure S2A,C). Nevertheless, unmeasured physiological variability inherent to enteric‐coated, acid‐labile formulations may still have contributed to the observed intra‐subject variability.

CONCLUSION

This study successfully demonstrated bioequivalence between the test and reference rabeprazole formulations using a 2 × 4 fully replicated crossover design and both the ABE and SABE approaches. Exploratory subgroup analyses suggested possible differences in systemic exposure according to CYP2C19 genotype and sex. However, these findings should be interpreted cautiously because the study was not prospectively powered for subgroup‐specific assessments, and CYP2C19 genotype and sex cannot directly explain within‐subject variability measured as intra‐CV in a replicated crossover design. Further adequately powered studies are needed to confirm the influence of CYP2C19 genotype and sex on rabeprazole pharmacokinetics.

CONFLICT OF INTEREST STATEMENT

Nakyong Lee is an employee of Daewoong Bio Corporation. The remaining authors declare no competing financial or non‐financial interests.

Supporting information

Supporting Information

DATA AVAILABILITY STATEMENT

All source data for this work (or generated in this study) are available upon reasonable request.