Exploration of suitable pharmacodynamic parameters for acarbose bioequivalence evaluation: A series of clinical trials with branded acarbose.

AIMS: To determine deficiencies in the Food and Drug Administration (FDA)'s guidance for assessing acarbose bioequivalence (BE) and to explore optimal pharmacodynamic (PD) metrics for better evaluation of acarbose BE.
METHODS: Three clinical trials with branded acarbose were conducted in healthy subjects, including a pilot study (Study I, n = 11, 50 and 100 mg), a 2×2 crossover BE study (Study II, n = 36, 100 mg) and a 4×4 Williams study (Study III, n = 16, 50/100/150 mg). Serum glucose concentrations were measured by the glucose oxidase method.
RESULTS: In Study I, compared with 50 mg acarbose, only 100 mg acarbose had a significantly lower Cmax0-4h than that of sucrose administration alone (7.96 ± 0.83 mmol/L vs 6.78 ± 1.02 mmol/L, P < .05). In Study II, the geometric mean ratios of the test formulation to the reference formulation (both formulations were the branded drug) for FDA PD metrics, ΔCmax0-4h and ΔAUC0-4h , were 0.903 and 0.776, respectively, and the 90% confidence intervals were 67.44-120.90 and 53.65-112.13, respectively. The geometric mean ratios (confidence interval) for possible optimal evaluation PD metrics (Cmax0-2h and AUC0-2h ) were 1.035 (94.23-112.68) and 0.982 (89.28-107.17), respectively. Further, Cmax0-2h and AUC0-2h also met the sensitivity requirements for BE evaluation in Study III.
CONCLUSION: Considering the mechanisms of action of acarbose, the PD effect was shown to be dose independent during the 2-4 hours postadministration of acarbose. Hence PD metrics based on the serum glucose concentration from 0 to 2 hours (Cmax0-2h and AUC0-2h ) are more sensitive than the FDA-recommended PD metrics for acarbose BE evaluation from 0-4 hours (ΔCmax0-4h and ΔAUC0-4h ). The trial has been registered at the Chinese Clinical Trial Registry (http://www.chictr.org.cn, ChiCTR1800015795, ChiCTR-IIR-17013918, ChiCTR-IIR-17011903). All subjects provided written informed consent before screening.

What is already known about this subject

The guidance issued by the Food and Drug Administration (FDA) helps the development of new drugs in many countries. Unfortunately, previous studies , , , , found that the FDA guidance for evaluating the bioequivalence (BE) of acarbose tablets appeared to be inappropriate in practice.

What this study adds

Compared with ΔCmax0–4h and ΔAUC0–4h proposed by the FDA, we proved that Cmax0–2h and AUC0–2h exhibited superior applicability and sensitivity in the evaluation of acarbose BE in healthy subjects. Cmax0–2h and AUC0–2h could be superior pharmacodynamic metrics for evaluating acarbose BE in the future.

INTRODUCTION

As the largest and among the most respectable drug review and regulatory agencies, the Food and Drug Administration (FDA) has a tremendous influence on drug development both in the USA and throughout the world. To improve the quality and efficiency of drug development, the FDA has developed and published many guidance documents to guide and accelerate the development of new drugs, making substantial contributions to innovation in drug development and to the availability of drug use globally. Therefore, the scientificity and feasibility of the published guidance have received a great deal of attention.

Acarbose is an alternative drug to metformin, which is the first‐line treatment drug for type 2 diabetes. , , As a prototypical α‐glucosidase inhibitor, acarbose acts on the brush border of small intestinal mucosal epithelial cells by reversibly binding to α‐glucosidase, competitively inhibiting the activity of the enzyme and delaying the breakdown of starch into glucose, thereby reducing the absorption of glucose in the intestine to reduce postprandial hyperglycaemia. ,

Because of the extremely low bioavailability of acarbose due to poor absorption in the gastrointestinal tract, the main site of acarbose action, bioequivalence (BE) testing of generic acarbose formulations cannot be based on a pharmacokinetic endpoint. In 2009, the FDA released a draft guidance for the BE assessment of acarbose tablets based on pharmacodynamic (PD) parameters and revised the guidance in 2017. The guidance recommended that generic products should be demonstrated to be therapeutically equivalent by conducting equivalence studies either in vitro or in vivo. The prerequisite for implementation of an in vitro BE study is that the generic product has exactly the same formulation as the branded product; however, the formulation of the branded product was undisclosed. Thus, the choice for most generic pharmaceutical companies is to conduct BE studies in vivo. Based on the FDA's guidelines, a 2‐way crossover design was used; generally, 75 g of sucrose was orally administered on Day 1 of each period (serum glucose baseline), followed by coadministration of 75 g sucrose/acarbose on Day 2. Blood samples were collected throughout the 4 hours after dosing for analysis of serum glucose concentration on Day 1 and Day 2. The alteration of the area under the serum curve (AUC; ΔAUC0–4h) and maximum plasma concentration (Cmax; ΔCmax0–4h) after acarbose administration vs baseline levels are considered to be effective PD metrics. Thus far, owing to the importance of the FDA's guidance in drug research and development, multiple countries have carried out acarbose BE studies with reference to these guidelines. , , , , However, both of these parameters are unsuitable for BE testing mainly because of the negative values of ΔAUC0–4h and ΔCmax0–4h during data processing, and the proportion of negative values can reach up to 30%. To guide BE studies, several new PD metrics have been proposed by different groups, , , , , but the applicability, sensitivity and scientificity of these new PD metrics still need further confirmation because the effects of the formulation on acarbose BE cannot be ruled out. Therefore, an exploration of scientific and reasonable PD metrics for evaluating the BE of acarbose tablets in vivo is necessary and highly sought after.

Using the branded drug, we conducted 3 studies, including a pilot study exploring the minimum appropriate dosage in 11 healthy volunteers at 2 doses of acarbose (50 and 100 mg), a 2×2 crossover BE study in 36 healthy volunteers at a single dose (100 mg), and a 4×4 Williams study in 16 healthy volunteers at 3 doses (50, 100 and 100 mg). The aim was to determine whether the FDA's guidance for assessing acarbose BE is appropriate and identify what the possible deficiencies are. We also attempted to explore optimal PD parameters by thorough analysis of the mechanism of action of acarbose together with related clinical data for better evaluation of acarbose BE in the future (Table 1).

TABLE 1

The objective and design of each study

	Study I (n = 12)	Study II (n = 36)	Study III (n = 16)
Objective	Preliminary dose exploration	PD metrics exploration for BE evaluation	Dose–response relationship and sensitivity of PD metrics
Study design	Randomized, 2×2 crossover design	Randomized, 2×2 crossover design	Randomized, 4×4 Williams design
Dose administration	Day 0 (baseline)	Day 1 (75 g of sucrose)	Day 1 (75 g of sucrose)
	Day 1 (75 g of sucrose)
			Day 2 (75 g of sucrose
			+50 mg (A) or 100 mg (B3) or 100 mg (B4) or 150 mg (C) acarbose
	Day2(75 g of sucrose	Day2(75 g of sucrose
	+50 mg (A) or 100 mg (B) acarbose)	+100 mg (B1) or 100 mg (B2) acarbose)
Sequence	AB, BA	B1B2, B2B1	AB3CB4, B3B4AC, B4CB3A, CAB4B3
Wash‐out day	7 wash‐out days	7 wash‐out days	7 wash‐out days
Sample collection times	0, 0.25, 0.5, 0.75, 1, 1.5, 2, 2.5, 3, 3.5, 4 h	0, 0.25, 0.5, 0.75, 1, 1.5, 2, 2.5, 3, 3.5, 4 h	0, 0.25, 0.5, 0.75, 1, 1.25, 1.5, 1.75, 2, 2.25, 2.5, 2.75, 3, 3.25, 3.5, 3.75, 4 h

Note: The letters A, B and C represent treatments A, B and C are acarbose doses of 50, 100 and 150 mg, respectively; the numbers after the same letter indicate the replicate treatments with same acarbose dose. The sample size mentioned in the table was the scheduled sample size.

METHODS

Study design

Three independent studies were included: Study I was a pilot study using a 2×2 crossover design, exploring the lowest significantly effective dose of acarbose tablets in a healthy Chinese population and providing a theoretical basis for dose optimization in subsequent studies. Study II was an applicability 2×2 crossover BE study using the same branded acarbose, assessing the current PD metrics for BE evaluation of acarbose tablets in healthy Chinese volunteers and defining which PD metrics recommended by the FDA apply to the Chinese population. Study III was a dose–response relationship analysis study using a 4×4 Williams design to evaluate the sensitivity of the PD metrics for acarbose tablets, verify the sensitivity of the PD metrics and assess whether a dose–response relationship exists. The detailed design for each study is elaborated in Table 2.

TABLE 2

Characteristics of healthy volunteers in Studies I, II and III

	Study I (n = 11)	Study II (n = 36)	Study III (n = 16)
Sex (male/female)	11/0	36/0	8/8
Age (y)	24 ± 3	25 ± 5	26 ± 6
Height (m)	1.71 ± 0.07	1.71 ± 0.06	1.63 ± 0.05
Weight (kg)	63.5 ± 5.0	64.3 ± 7.4	57.0 ± 7.9
BMI (kg/m2)	21.7 ± 2.0	22.0 ± 2.2	21.3 ± 2.0

Note: Values of age, height, weight and body mass index (BMI) are the mean ± standard deviation. Twelve subjects participated in Study I, of whom 11 completed the study (1 individual withdrew from the study due to personal reasons was unrelated to the study); 36 subjects participated in and completed Study II; and 16 subjects participated in and completed Study III. The sample size mentioned in the table was the actual sample size.

For all the studies, eligible subjects were selected from healthy Chinese male or female volunteers aged ≥18 years with a body weight ≥ 50 kg and body mass index 19–26 kg/m2. A full medical examination was performed on all subjects including a physical examination, medical history, biochemical tests and electrocardiogram. An oral glucose tolerance test was also performed to ensure normal oral glucose tolerance. Subjects were admitted to the clinical trial centre on the day before the first treatment for each period. Subjects maintained alimentary abstinence for over 10 hours prior to each treatment and continued for 4 hours after dosing. Diet and exercise were strictly controlled, and any excessive exertion or lying supine for long periods were not allowed.

We assure that the work described has been carried out in accordance with The Declaration of Helsinki. The study was approved by the medical ethics committee of the Third Xiangya Hospital of Central South University, and the ethical approval number was R17005.

Study drug

Acarbose tablets: Bayer; main components: acarbose; specifications: 50 mg per tablet; batch number: BJ34593; usage: oral; dose: a single dose according to the protocol.

Sample processing and detection

After the blood sample was collected, it was mixed 5 or 6 times. After coagulation, the blood was centrifuged at 2301 g at 4°C for 10 minutes, and the upper serum layer was transferred to a new tube and stored at 2 to 8°C for use. The backup serum tubes were pre‐frozen, stored at −20 ± 5°C and transferred to −80°C at the end of each cycle. The serum samples were sent to the clinical laboratory of the Third Xiangya Hospital of Central South University for analysis by the glucose oxidase method after the end of the daily testing, and the backup serum samples, if needed, were measured within 48 hours.

The serum glucose concentration in this research was measured by the glucose oxidase method using a Hitachi 7170 automatic analyser (Hitachi Co., Tokyo, Japan) in central laboratory of the Third Xiangya Hospital of Central South University, according to the standard laboratory manual for glucose determination. All results of glucose concentration were obtained with the same automatic analyser. Quality control was performed on the automatic analyser at the beginning and the end of the glucose testing using the manufacturer's control solution according to the manufacturer's instructions to ensure ongoing accuracy. Two levels of control (known normal and abnormal glucose values) were run to monitor the validity of the reaction. The precision and accuracy of quality control should state a coefficient of variation <5%.

Statistical analysis

BE candidate PD metrics calculation

All candidate BE PD metrics were analysed using R software (version number: 3.3.2). In our research, the ΔC max0–4h was the maximum difference between the baseline glucose profile from 0 to 4 hours determined on the day prior to drug treatment (Day 1) and the glucose profile from 0 to 4 hours determined on the day of drug treatment (Day 2), ΔC max0–4h = Cmax0–4h, Day1 – Cmax0–4h, Day2; the ΔAUC0–4h was the difference in postprandial area under the curve from 0 to 4 hours (AUC0–4h) for blood glucose following sucrose load without (Day 1) and with acarbose (Day 2), ΔAUC0–4h = AUC0–4h, Day1 ‐ AUC0–4h, Day2. AUC0–2h and Cmax0–2h: the AUC and Cmax of the glucose profile after each time point minus the baseline glucose concentration (0 h) on the day of drug treatment (Day 2). The other 14 PD metrics are detailed in the supplement.

BE evaluation

The BE evaluation was performed using Phoenix WinNonlin software (version number: 6.1) in Study II for BE parameter exploration and Study III for BE parameter sensitivity. When the 90% confidence interval of the ratio of the geometric mean of the PD metrics was between 80.00 and 125.00%, the 2 sets of parameters were considered equivalent. The data from Study II were used to screen for candidate PD metrics that met the BE criteria. The sensitivity of each candidate was evaluated by equivalence testing of parameters between different dose groups in Study III (e.g. 50 mg [A] vs 100 mg [B3]; 50 mg [A] vs 100 mg [B4]; 50 mg [A]vs 150 mg [C]; 100 mg [B3] vs 150 mg[C]; and 100 mg [B4] vs 150 mg [C]). Note that, according to actual conditions, negative concentrations were considered to be zero when calculating the 2 PD metrics Cmax0–2h and AUC0–2h. For the 2 FDA‐recommended parameters, ΔCmax0–4h and ΔAUC0–4h, negative values were not included in the BE analysis.

Dose–effect relationship of PD metrics

For a PD endpoint BE study, the FDA recommends that the dose should be on the linear portion of the PD dose–response curve. Therefore, the linear dose–response relationship of PD metrics was evaluated in R by using the lm() function.

Nomenclature of targets and ligands

Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY.

RESULTS

Study population

Twelve subjects participated in Study I, of whom 11 completed the study ( 1 individual withdrew from the study due to personal reasons unrelated to the study); 36 subjects participated in and completed Study II; and 16 subjects participated in and completed Study III. The demographic characteristics of subjects are detailed in Table 2. Subjects who completed the study per protocol were included in the statistical analysis.

Postprandial high‐sugar model and acarbose hypoglycaemic effect in the human body

In the pilot study, we initially explored the effective dose of acarbose tablets. The ΔCmax0–4h of coadministration of 50 mg of acarbose/sucrose was (0.42 ± 1.13) mmol/L, and no significant difference in ΔCmax0–4h was found between subjects receiving acarbose tablets vs control at a dose of 50 mg (P > .05). However, the ΔCmax0–4h of 100 mg acarbose was (1.17 ± 0.79) mmol/L, which was much higher than the ΔCmax0–4h of 50 mg (Figure 1). Therefore, 100 mg of acarbose showed a significant hypoglycaemic effect compared to 50 mg and should subsequently be used as the optimal dose.

FIGURE 1

Serum glucose concentration–time curve of the administration of 75 g sucrose in the absence or presence of acarbose. Serum glucose concentration are represented as mean ± standard deviation and shown with the error bars. Jittering were used to avoid overlapping of the points and whiskers. Day 0: no administration of sucrose or acarbose; day 1: administration of sucrose (75 g); day 2: sucrose in addition of acarbose. The treatments in terms of A, B (B1, B2) and C represent that the investigated dose acarbose in each treatment arm was 50, 100 and 150 mg, respectively, and B1 and B2 arm indicate the replicate treatments with same acarbose dose. (A, B) Serum glucose profiles receiving treatments A and B in Study I, which was a preliminary dose‐exploration study using a 2×2 crossover design (AB and BA). (C) Serum glucose concentration–time curves receiving the treatment B1 and B2 in study II for the purpose of pharmacodynamic metric exploration with a 2×2 crossover design (B1B2 and B2B1). (D) Serum glucose profiles of receiving acarbose from 50 mg to 150 mg in study III, which adopted a 4×4 Williams design (AB3CB4, B3B4AC, B4CB3A and CAB4B3). To show the dose–response relationship of acarbose, 2 groups (B3, B4) were combined in (D)

In Study I, the profiles of serum glucose levels from 0 to 4 hours on Day 0 without administration of sucrose or acarbose were essentially the same, with no significant differences between the 2 periods (P > .05), indicating that healthy volunteers had fewer fluctuations in serum glucose during fasting (Figure 1). The curve of serum glucose concentration over time after sucrose administration alone was revealed to be an approximate sinusoid. During the 2 hours postdosing, serum glucose levels rapidly rose to the highest level, followed by a continuous decrease to even lower than baseline, but finally returned to the initial baseline level between 2 and 4 hours (Figure 1). In contrast, the extent of serum glucose reduction was less and was almost stable near the baseline level in the 2–4 hours after coadministration of acarbose and sucrose. Since Studies I and II were densely sampled only during the initial hour, Study III was conducted and intensively sampled the entire 4 hours period to better characterize the curve, and the curves for these 3 studies demonstrated similar features (Figure 1D).

Defects of the current PD metrics proposed by the FDA

Guidance published by the FDA requires logarithmic transformation of ΔAUC0–4h and ΔCmax0–4h before performing BE analysis, but we found that 11.1% of the ΔAUC0–4h values were negative and thus impossible to convert logarithmically. We evaluated the applicability of the 2 FDA PD metrics (ΔCmax0–4h and ΔAUC0–4h) with B1 as the reference formulation and B2 as the test formulation (both B1 and B2 were the branded product); both ΔCmax0–4h and ΔAUC0–4h failed to meet the BE criteria (as shown in Table 3). In addition, Study III (the dose–effect relationship of acarbose) showed that the concentration–time curves from 2 to 4 hours nearly overlapped after administration of 50, 100 or 150 mg acarbose, and the effect of acarbose is not dose dependent from 2 to 4 hours. Further, ΔAUC0–4h recommended by the FDA has not significantly different between different doses of acarbose (P = .588, Figure 2). Therefore, the PD metrics proposed by the FDA are defective and unable to evaluate acarbose BE in humans.

TABLE 3

Bioequivalence evaluation of 2 pharacodynamic (PD) metrics recommend by the Food and Drug Administration (ΔCmax0–4h and ΔAUC0–4h) based on serum glucose concentrations for study II (n = 36)

PD metrics	Mean ± SD		Ratio	90% CI	CV%
	B1 (n = 36)	B2 (n = 36)
ΔCmax0–4h (mmol/L)	1.56 ± 0.14	1.42 ± 0.15	90.30	(67.44,120.90)	82.36
ΔAUC0–4h (h*mmol/L)	1.56 ± 0.23	1.35 ± 0.18	77.56	(53.65,112.13)	106.23

ΔC max0–4h: the maximum difference between the baseline glucose profile from 0 to 4 hours determined on the day prior to drug treatment (Day 1) and the glucose profile from 0 to 4 hours determined on the day of drug treatment (Day 2). ΔC max0–4h = Cmax0–4h, Day1 ‐ Cmax0–4h, Day2; ΔAUC0–4h: the difference in postprandial area under the curve from 0 to 4 hours (AUC0–4h) for blood glucose following sucrose load without (Day 1) and with acarbose (Day 2). ΔAUC0–4h = AUC0–4h, Day1 ‐ AUC0–4h, Day2.

SD, standard deviation; CI, confidence interval; CV, coefficient of variance

FIGURE 2

Dose–effect relationship of Food and Drug Administration pharmacodynamic (PD) metrics (ΔCmax and ΔAUC0–4) and recommended PD metrics (Cmax0_2h and AUC0–2h). The linear dose–response relationship of PD metrics was evaluated in R using the lm() function. The black dots represent each calculated specific PD metric. The blue lines mean the linear regression lines and the shadow parts represent the 95% confidence interval around the regression lines. P < .05 means significant linear dose–response relationship of PD metrics. RCMD: recommended

PD metric candidates

A total of 18 PD metrics was evaluated for applicability and sensitivity of acarbose BE study. Some PD metrics were selected according to the FDA guidance and literatures, and the remains were first proposed based on the mechanisms of action of acarbose. The applicability and sensitivity were evaluated based on the results of Study II (refer to Table S1) and Study II (refer to Table S1), respectively, and summarized in Table S2. Eleven PD metrics satisfied the criteria of applicability evaluation; however, only Cmax0_2h, AUC0–2h and ΔAUC0–2 satisfied the further sensitivity evaluation. Some PD metrics were calculated by ratio method (RatioCmax and RatioAUC0–4), or without considering the baseline (Cmax‐Day2 and AUC0–4‐Day2). Obviously, these approaches were too insensitive to be considered as the PD metric candidates for BE of acarbose. In addition, several PD metrics have only satisfied the sensitively evaluation for some but not all groups (Av‐ ΔAUC, GE‐Day2, GE’‐Day2 and fAUC‐Day2), and also cannot be considered to be appropriate PD metric candidates. Taken together, we proposed Cmax0_2h and AUC0–2h as the optimal PD metrics to evaluate the BE of acarbose. Table 4 presents that these parameters met the applicability and sensitivity requirements for BE evaluation.

TABLE 4

Bioequivalence and sensitivity analysis of Cmax0_2h and AUC0–2h calculated based on serum glucose concentrations

Study	Data	C max0–2h (mmol/L)			AUC0–2h (h*mmol/L)
		Ratio	90% CI (%)	CV%	Ratio	90% CI (%)	CV%
Study I (n = 11)	50 (A)‐100 (B) b	150.90	(127.71–178.29)	21.49	185.88	(157.08–219.96)	21.70
Study II (n = 36)	100 (B1)‐100 (B2) a	103.45	(94.23–112.68)	20.52	98.23	(89.28–107.17)	22.35
Study III (n = 16)	50 (A)‐100 (B3) b	107.98	(90.92–128.24)	26.29	100.60	(75.80–133.51)	41.43
Study III (n = 16)	50 (A)‐100 (B4) b	108.58	(89.66–131.49)	31.48	111.74	(92.91–134.37)	30.28
Study III (n = 16)	50 (A)‐150 (C) b	135.17	(112.19–162.85)	30.60	161.61	(126.85–205.90)	40.40
Study III (n = 16)	100 (B3)‐150 (C) b	79.50	(65.72–96.18)	31.30	58.99	(45.56–76.38)	40.27
Study III (n = 16)	100 (B4)‐150 (C) b	80.33	(68.79–93.80)	25.28	67.56	(52.05–87.69)	30.54
Study III (n = 16)	100 (B3)‐100 (B4) a	98.97	(81.62–120.02)	31.71	90.96	(69.87–118.41) c	44.32

Applicability analysis;

Sensitivity analysis; the 90% confidence interval (CI) of the ratios for Cmax0–2h and AUC0–2h falling within 80.00–125.00% was considered indicative of equivalence for 2 formulations. C max0–2h: the maximum glucose concentration of the glucose profile after each time point minus the baseline glucose concentration (0 h) on the day of drug treatment (Day 2); AUC0–2h: the postprandial area under the curve from 0 to 2 hours for blood glucose after each time point minus the baseline glucose concentration (0 h) on the day of drug treatment (Day 2).

The AUC0–2h of 100 (B3)‐100 (B4) in Study III failed for applicability analysis, this may largely be attributed to the small sample size.

CV, coefficient of variance

DISCUSSION

We conducted an optimal dose‐exploration study, a BE study and a 3‐dose self‐controlled study in healthy volunteers using branded acarbose, and the data indicated that the FDA‐recommended PD metrics (ΔCmax0–4h and ΔAUC0–4h) for acarbose BE evaluation in vivo are defective in terms of both mechanism and practice. The mechanism of action of acarbose and observations of the activity of acarbose at 2–4 hours were basically consistent among different doses, which was indicated by the similar serum glucose response curves of the 50, 100 and 150 mg doses at the later time point after coadministration of sucrose and acarbose (e.g. 2–4 h). These results led us to focus on the other 2 parameters, Cmax0–2h and AUC0–2h, based primarily on the 0–2 hours effect of acarbose, which are relatively optimal PD metrics for the evaluation of acarbose BE in vivo.

The main reasons for the deficiencies in FDA parameters in acarbose BE evaluation

In comparison with sucrose alone, the serum glucose concentration–time curve in most of the healthy volunteers after the simultaneous administration of sucrose and acarbose tablets showed a significant reduction in serum glucose peak, along with an increase in the serum valley, presenting a trend of an initial increase followed by a dramatic decrease to the approximate baseline levels (Figure 1). This phenomenon could be partially explained by the mechanism of action of acarbose. Acarbose binding to α‐glucosidase leads to a reduction in carbohydrate absorption in the upper part of the duodenum and jejunum, resulting in a decreased peak blood glucose value compared to that of sucrose administration. Thereafter, unabsorbed carbohydrates would be absorbed in the lower and middle parts of the small intestine or even the colon, and then slightly increase valley blood glucose values. In addition, the administration of acarbose attenuates a rapid increase in blood glucose concentrations in response to a corresponding decrease in insulin secretion, which may somehow prevent and control a possible overreaction of insulin in healthy subjects. This phenomenon not only is consistent with previous reports but also can explain other clinical experiences, such as 3 hours after having a carbohydrate‐based breakfast in the morning, a group of people have experienced extreme hunger for approximately 30 minutes.

The main reasons for the deficiencies in FDA parameters in acarbose BE evaluation in vivo may include the following. First, due to the negative feedback regulation of serum glucose itself, the serum glucose concentration was lower 2–4 hours after sucrose administration alone than at the baseline level, whereas after coadministration of sucrose and acarbose, the serum glucose values from 2 to 4 hours were similar to the baseline levels. As a result, some negative values of ΔAUC0–4h are present and cannot be analysed through logarithmic conversion. Second, as the effect of acarbose is not dose dependent from 2 to 4 hours (Figure 1D), prolongation of the blood sample collection time to 4 hours for PD analysis partially diluted the PD effect that occurs during the first 2 hours, reducing the sensitivity of the evaluation. Thus far, the European Medicines Agency and Pharmaceuticals and Medical Devices Agency have not yet released acarbose‐specific BE guidance. However, in the overview of the marketing application information of acarbose, we found that some acarbose BE studies have used the ratio method of PD parameters for BE evaluation when applying to the European Medicines Agency market, and some studies have used baseline‐corrected PD parameters when applying to the Pharmaceuticals and Medical Devices Agency market.

The applicability and sensitivity of PD metrics

In fact, the findings of several recent acarbose BE studies , , , , performed in different countries indicated challenges in applying the current FDA‐recommended BE evaluation method, and some researchers attempted to explore and propose new PD metrics, such as PD metrics reflecting the fluctuation of blood glucose. However, all the PD metric exploration was based on the assumption that the reference and test formulations were bioequivalent, while in fact the premise was uncertain, so the applicability, sensitivity and scientificity of these new PD metrics still need further confirmation. Considering these limitations, our research was the first to determine the appropriateness of the FDA's guidance for assessing acarbose BE using branded drugs and to apply a more systematic assessment method to explore optimal PD metrics. The current study analysed the applicability and sensitivity of 18 PD metrics, most of which met the requirements for BE evaluation but did not show significant dose‐dependent differences at doses of 50, 100 and 150 mg, making it impossible to distinguish the differences in hypoglycaemic efficacy between different treatment groups (50 vs 100 mg or 100 vs 150 mg), and indicating insufficient sensitivity of these PD metrics (Tables S1 and S2, Figure S3).

In our research, Study I was a preliminary dose exploration that showed that healthy volunteers had fewer fluctuations in serum glucose during fasting without administration of sucrose or acarbose, and the fluctuations in serum glucose after administration of 75 g sucrose alone have no significance between periods. Compared to 50 mg, 100 mg of acarbose showed a significant hypoglycaemic effect and 100 mg was selected as the optimal dose for Study II. Study II was a PD metrics exploration for BE evaluation using the repeated 100 mg branded acarbose. The dose selection of Study III was based on the data of Studies I and II, and 150 mg was added to further verify whether 100 mg had reached the dose saturation point. To characterize a more remarkable profile of serum glucose level, the PD sampling were designed denser compared to Study I and Study II. The 2 100 mg treatments designed in Study III could also verify the rationale of Study II. In Study III, some PD metrics were applicable with 100 (B4) but not with 100 (B3). This may be attributed that the PD metrics were calculated based on the glucose concentrations which affected by a number of factors, such as the intrinsic and extrinsic factor. Here, the Study III has a relative long clinical trial span (24 days) that could increase the intrasubject variability between periods. Since the PD effect was dose independent during the 2–4 hours period postadministration of acarbose, we speculate that the prolongation of blood sample collection time to 4 hours may partially eliminate the PD effects in the first 2 hours and result in a reduction in sensitivity. Therefore, we recommended 2 new PD metrics: Cmax0–2h and AUC0–2h, which are baseline corrected PD parameters obtained from analysis of blood samples collected between 0 and 2 hours after administration of sucrose together with acarbose. Through the correction of the zero point, lowering the blood sugar value (i.e. Cmax0–2h) and area under the serum curve (i.e. AUC0–2h) increases the sensitivity of the parameters and shows a good dose–effect relationship. The accepted criterion for showing BE for Cmax0–2h and AUC0–2h is that the 90% confidence interval of the B1/B2 ratio falls within 80.00–125.00%, while ensuring that all of the primary PD parameters of the generic product as well as the branded product show statistically significant differences compared to sucrose alone.

The potential factors of individual differences in response to acarbose treatment

Individual differences in response to acarbose treatment were observed; 13% of the subjects failed to exhibit hypoglycaemic effects after administration of acarbose tablets. To our knowledge, no related literature or relevant data associated with genetic polymorphisms encoding glycosidase genes have been reported. In contrast, several groups , , focused on the interaction between acarbose and the intestinal flora. Gu et al found that acarbose significantly changed the relative abundance of some of the intestinal microbial genes and mOTUs in the gastrointestinal tract with regard to the improvement in fasting serum glucose, insulin secretion, C‐peptide levels and insulin resistance. The efficacy of acarbose was more intensive in patients with type B diabetes than in those with type P diabetes. Hereby, we conducted a preliminary intestinal microbiological study through the analysis of faecal samples and found differential expression of 8 intestinal flora between 2 different hypoglycaemic groups (Figure S1, S2, Table S3). Therefore, we speculated that differential expression of intestinal flora among individuals may be an important factor leading to the variation in efficacy of acarbose tablets.

Limitations

Potential limitations exist in this study. First, due to the absence of the specifications of 25‐mg acarbose tablets, the dose–effect relationship of acarbose (indicating the sensitivity of the PD metrics) was specified at 3 doses, 50, 100 or 150 mg. We supposed that screening for more sensitive PD metrics with acarbose 75, 100 and 125 mg would be a good choice. Second, the negative concentration considered to be zero for calculating the 2 PD metrics (Cmax0–2h and AUC0–2h) could create a certain degree of bias. However, compared with the baseline glucose concentration level (5.02 ± 0.23 mmol/L), the proportion of absolute values of negative concentrations from 0 to 2 hours (0.43 ± 0.35 mmol/L) to the baseline was approximately 8.6%. Therefore, the possible bias caused by negative concentrations could have a minor effect on the result of PD metrics. Third, in our research, there was no strict sample size calculation criteria. Study I was a pilot trial, a small sample size (n = 12) was adequate to achieve the purpose. The sample size of Study II was calculated based on the results of Study I as well as literature. Based on the purposes of Study III, a relatively small sample size could better reflect the sensitivity of PD metrics. Comprehensive consideration of the 4×4 Williams design, 16 subjects was finally selected. Additionally, as the studies were conducted only in a Chinese population, the outcome may be different between different populations.

CONCLUSIONS

Considering the mechanisms of action of acarbose, the PD effect was shown to be dose independent manner during the 2–4 hours period postadministration of acarbose, so the PD metrics calculated based on the serum glucose concentration from 0 to 2 hours are more sensitively than that derived from 0–4 hours. Meanwhile, The FDA‐recommended PD metrics (ΔCmax0–4h and ΔAUC0–4h) for acarbose BE evaluation in vivo are insufficient in terms of both mechanism and practice. Similarly, the PD metrics calculated by ratio method or without considering the baseline were too insensitive to be considered as the PD metric candidates. Collectively, Cmax0–2h and AUC0–2h could be optimal PD metrics for evaluating acarbose BE in the future.

CONTRIBUTORS

The principal investigator of the three clinical trials was Dr. Guo‐ping Yang. Guo‐ping Yang, Jin‐bo Yang and Min Li conceived and designed the study; Jie Huang, Wen‐yu Liu, Jing‐jing Yu, Xiao‐yan Yang, Shuang Yang, Jin‐lian Xie and Zhi‐jun Huang implemented the clinical trials; Hui Chen and Cheng‐xian Guo assessed the biological samples; Qi Pei, Jie Huang, Wen‐yu Liu and Jing‐jing Yu analysed and interpreted the data; Jie Huang, Wen‐yu Liu and Chan Zou wrote the paper; Guo‐ping Yang and Qi Pei reviewed the paper.

COMPETING INTERESTS

There are no competing interests to declare.

FULL DISCLOSURE

We confirm that the manuscript has not been previously published in any language anywhere and that it is not under simultaneous consideration by another journal.

FIGURE S1 Group comparison analysis: evolutionary branch graph

FIGURE S2 Group comparison analysis: linear discriminant analysis value distribution histogramTABLE S1 The result of the applicability of pharmacodynamic metrics calculated based on glucose concentration from the data of Study II (n = 36)

FIGURE S3 The result of sensitivity of pharmacodynamic metrics calculated based on glucose concentration from the data of Study III. The solid dot and the solid line represented the ratio and 90% confidence interval of bioequivalence evaluation, respectively. The 2 dotted lines in each graph represented the 80.00 and 125.00% limits.TABLE S2 The applicability and sensitivity of the pharmacodynamic metrics calculated based on glucose concentration for bioequivalence evaluation.TABLE S3 Differential bacteria between effective and invalid groups (P < .05)

Click here for additional data file.