Analysis of Intra- and Intersubject Variability in Oral ... - ACS Publications

Article pubs.acs.org/molecularpharmaceutics

Analysis of Intra- and Intersubject Variability in Oral Drug Absorption in Human Bioequivalence Studies of 113 Generic Products Masahisa Sugihara,*,† Susumu Takeuchi,† Masaru Sugita,† Kazutaka Higaki,‡ Makoto Kataoka,§ and Shinji Yamashita§ †

Sawai Pharmaceutical Co., Ltd., 5-2-30 Miyahara, Yodogawa-ku, Osaka, Osaka 532-0003, Japan Faculty of Pharmaceutical Sciences, Okayama University, 1-1-1 Tsushima-naka, Kita-ku, Okayama 700-8530, Japan § Faculty of Pharmaceutical Sciences, Setsunan University, 45-1 Nagaotoge-cho, Hirakata, Osaka 573-0101, Japan ‡

ABSTRACT: In this study, the data of 113 human bioequivalence (BE) studies of immediate release (IR) formulations of 74 active pharmaceutical ingredients (APIs) conducted at Sawai Pharmaceutical Co., Ltd., was analyzed to understand the factors affecting intra- and intersubject variabilities in oral drug absorption. The ANOVA CV (%) calculated from area under the time−concentration curve (AUC) in each BE study was used as an index of intrasubject variability (Vintra), and the relative standard deviation (%) in AUC was used as that of intersubject variability (Vinter). Although no significant correlation was observed between Vintra and Vinter of all drugs, Vintra of class 3 drugs was found to increase in association with a decrease in drug permeability (Peff). Since the absorption of class 3 drugs was rate-limited by the permeability, it was suggested that, for such drugs, the low Peff might be a risk factor to cause a large intrasubject variability. To consider the impact of poor water solubility on the variability in BE study, a parameter of Peff/Do (Do; dose number) was defined to discriminate the solubility-limited and dissolution-rate-limited absorption of class 2 drugs. It was found that the class 2 drugs with a solubility-limited absorption (Peff/Do < 0.149 × 10−4 cm/s) showed high intrasubject variability. Furthermore, as a reason for high intra- or intersubject variability in AUC for class 1 drugs, effects of drug metabolizing enzymes were investigated. It was demonstrated that intrasubject variability was high for drugs metabolized by CYP3A4 while intersubject variability was high for drugs metabolized by CYP2D6. For CYP3A4 substrate drugs, the Km value showed the significant relation with Vintra, indicating that the affinity to the enzyme can be a parameter to predict the risk of high intrasubject variability. In conclusion, by analyzing the in house data of human BE study, low permeability, solubility-limited absorption, and high affinity to CYP3A4 are identified as risk factors for high intrasubject variability in oral drug absorption. This information is of importance to design the human BE study for oral drug products containing APIs with a risk of large intrasubject variability in oral absorption. KEYWORDS: bioequivalence, intrasubject variability, BCS, solubility-limited absorption, CYP3A4

■

INTRODUCTION

tested product. Major causes of the intrasubject variability include variabilities in the absorption and the pharmacokinetic profiles of the API as well as those in the bioperformance of the formulation. For conducting a BE study of a highly variable drug product, the EMA and FDA recommend a referencescaled average BE (RSABE) approach.4 In the RSABE approach, the reference product is administered twice in the study and the acceptance limits scale is based on the intrasubject variability of the reference product. In Japan, BE can be proven if differences in the geometric mean ratios are within the range of log(0.90)−log(1.11), but only if the outcomes of a dissolution test are similar for the reference and test products.

Bioequivalence (BE) study is a test to compare the bioavailability of the active pharmaceutical ingredient (API) for ensuring the therapeutic equivalence of a given pair of different formulations containing the same API. A BE study is usually carried out in a 2-period, 2-sequence, crossover design, in which subjects are administered the reference and the test formulations alternately. By comparing the area under the time−concentration curve (AUC) and the maximal concentration (Cmax) of the blood levels of the API, two formulations are considered to be bioequivalent if the 90% confidence intervals of differences in the average values of logarithmic parameters are within the acceptable range of log(0.80)− log(1.25).1−3 Both intersubject and intrasubject variabilities are observed in BE studies. Because intersubject variability is not pertinent to crossover study design, high intrasubject variability often causes problems in a BE study, making it difficult to prove the BE of a © XXXX American Chemical Society

Received: August 5, 2015 Revised: October 21, 2015 Accepted: November 15, 2015

A

DOI: 10.1021/acs.molpharmaceut.5b00602 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

B

mosapride citrate nilvadipine

itopride hydrochloride limaprost alfadex losartan potassium

imidapril hydrochloride

glimepiride

fluvoxamine maleate

fexofenadine hydrochloride fluvastatin sodium

epinastine hydrochloride

doxazosin mesilate

clotiazepam donepezil hydrochloride

cibenzoline succinate

carvedilol cetirizine hydrochloride

brotizolam cabergoline

benazepril hydrochloride beraprost sodium betaxolol hydrochloride

atorvastatin calcium

alprazolam amlodipine besilate

compound

0.8 mg 2.5 mg 5 mg 5 mg 10 mg 5 mg 20 μg 5 mg 10 mg 0.25 mg 0.5 mg 1 mg 2.5 mg 5 mg 10 mg 50 mg 100 mg 5 mg 5 mg 10 mg 1 mg 2 mg 10 mg/g 20 mg 60 mg

20 mg 30 mg 25 mg 50 mg 75 mg 1 mg 3 mg 2.5 mg 5 mg 10 mg 50 mg 5 μg 50 mg 100 mg 5 mg 2 mg

FCT FCT FCT FCT FCT UCT UCT UCT UCT UCT FCT UCT FCT FCT FCT FCT

strength

UCT FCT FCT FCT FCT UCT FCT FCT FCT ODT UCT UCT FCT ODT FCT FCT FCT FCT FCT FCT UCT UCT DS FCT FCT

dosage forma

3.436 3.038

2.65 3.44 4.112

1.531

3.96

3.321

4.048

1.96

3.506

2.48

4.11 4.6

3.07

4.041 2.08

2.358 4.171

1.824 2.044 2.8

4.457

2.205 3.434

cLogP

2.73 1.84

1.75 1.38 0.87

0.35

0.27

2.19

2.51

0.59

2.72

1.39

7.45 3.42

3.6

1.43 0.94

5.72 1.15

0.53 2.05 1.96

1.73

5.52 0.71

Peff (× 10−4cm/s)

Table 1. List of 113 Oral Drug Products of 74 APIs

0.05 0.1

> 1000 500 336

91

0.102

41.7

47.3

1.98

100

6.67

0.05 125

23.75

0.013 > 1000

0.02 0.2

128 100 > 1000

0.15

0.08 2.22

Solubility (mg/mL)

0.0028 0.0042 0.0040 0.0080 0.012 0.065 0.196 0.00018 0.00037 0.00073 0.00033 0.000000067 0.0010 0.0020 0.667 0.133

0.067 0.0075 0.015 0.222 0.444 0.00026 0.0000027 0.000033 0.000067 0.083 0.017 0.033 0.253 0.000033 0.000067 0.014 0.028 0.667 0.00027 0.00053 0.0010 0.0020 0.00033 0.0013 0.202

Do

890 594 548 274 183 4.13 1.38 1911 956 478 5250 20700000 877 438 4.10 13.8

82.8 94.6 47.3 7.79 3.89 2035 768750 58800 29400 68.6 69.0 34.5 5.64 28200 14100 257 128 11.2 12825 6413 1391 695 8160 2040 2.92

Peff/Do ( ×10−4cm/s)

1 1

1 1 1

1

1

1

1

1

1

1

1 1

1

1 1

1 1

1 1 1

1

1 1

BCS class

20 30 25 50 75 1 3 2.5 5 10 50 0.005 50 100 5 2

0.8 2.5 5 5 10 5 0.04 5 10 0.25 0.5 1 2.5 5 10 50 100 5 5 10 1 2 5 20 60

dose (mg)

22.6 21.9 9.5 9.8 10.4 8.2 6.4 28.5 33.1 20.1 5.4 25.2 11.6 11.4 21.9 19.4

5.0 10.7 6.0 13.7 8.9 16.2 35.6 5.3 4.0 11.2 25.6 21.5 16.0 6.2 4.7 4.7 4.2 7.3 4.7 4.4 17.4 19.8 14.4 21.6 32.9

AUC Vintra

38.8 32.9 52.8 94.3 47.4 26.1 20.4 36.1 49.0 39.1 25.1 36.8 36.6 36.7 51.6 50.5

18.5 23.2 19.4 50.1 44.4 27.3 41.7 15.9 23.8 28.9 36.1 35.8 38.6 17.9 15.1 21.7 17.4 35.8 16.9 18.0 36.2 36.7 31.7 23.3 36.7

AUC Vinter

3A4

2D6

3A4

3A4

3A4

3A4 and 2D6

3A4 3A4

3A4

3A4 3A4

CYP isoform

584

595

33

575

Km (μM)

16]

14

16

16

ref

Molecular Pharmaceutics Article


C

meloxicam

ecabet sodium etodolac itraconazole loratadine

ebastine

bicalutamide cefditoren pivoxil cilnidipine clarithromycin

amiodarone hydrochloride benidipine hydrochloride

ticlopidine hydrochloride toremifene citrate trandolapril zolpidem tartrate

temocapril hydrochloride

tandospirone citrate

tamoxifen citrate

risperidone

propiverine hydrochloride

pramipexole hydrochloride pravastatin sodium

paroxetine hydrochloride

compound

Table 1. continued

UCT UCT

5 mg 10 mg

2 mg 4 mg 8 mg 80 mg 100 mg 10 mg 50 mg 200 mg 5 mg 10 mg 0.667g/g 200 mg 50 mg 10 mg

FCT FCT FCT FCT FCT FCT FCT FCT FCT FCT GR FCT CAP UCT

5 mg 10 mg 10 mg 20 mg 0.5 mg 1 mg 2 mg 3 mg 20 mg

UCT UCT FCT FCT ODT FCT FCT FCT FCT

5 mg 10 mg 2 mg 4 mg 100 mg 40 mg 1 mg 5 mg 10 mg 100 mg

4 mg 10 mg 20 mg 0.125 mg

FCT FCT FCT UCT

FCT FCT UCT UCT FCT UCT UCT FCT FCT UCT

strength

dosage forma

2.292

4.93 3.52 6.046 5.051

7.062

2.707 2.711 5.54 1.97

5.709

9.13

4.388 6.53 4 2.826

2.102

1.5

6.818

2.711

3.874

2.048

1.76

4.24

cLogP

2.52

2.23 3.06 1.85 7.61

2.72

0.99 0.81 1.1 0.32

1.04

3.12

9.52 5.52 0.51 6.94

0.48

2.98

5.62

3.35

3.27

1.07

1.35

2.43


0.0055

0.01 0.11 0.0073 0.00088

0.0021

0.0044 0.01 0.000024 0.01

0.01

0.43

78 0.53 0.77 8.9

0.2

25

0.316

0.09

167

> 1000

> 1000

8.64

Solubility (mg/mL)

6.072 12.143

1.333 2.667 5.333 121.2 66.667 2778 33.3 133.3 15.80 31.60 666.7 12.12 45.475 75.758

0.0013 0.0027 0.067 0.133 0.009 0.507 0.0087 0.0037 0.007 1.540

0.000033 0.000067 0.00040 0.00080 0.037 0.074 0.148 0.222 0.422

0.267 0.0077 0.015 0.00000083

Do

0.42 0.21

0.78 0.39 0.20 0.0082 0.012 0.00040 0.0096 0.0024 0.17 0.086 0.0033 0.25 0.041 0.10

2235 1118 7.20 3.60 1114 10.9 58.8 1853 926 2.03

32100 16050 8191 4096 90.5 45.2 22.6 15.1 13.3

6.90 315 157 1620000


2

2 2 2 2

2

2 2 2 2

2

2

1 1 1 1

1

1

1

1

1

1

1

1

BCS class

5 10

2 4 8 80 100 10 50 200 5 10 1000 200 50 10

5 10 2 4 100 40 1 5 10 100

5 10 10 20 0.5 1 2 3 20

4 10 20 0.125

dose (mg)

6.6 7.0

36.2 18.2 20.4 25.4 30.2 23.1 20.7 14.1 13.6 16.7 23.2 9.2 44.9 10.5

18.4 15.3 9.0 9.6 20.6 8.8 12.3 19.1 11.3 11.9

60.7 29.5 20.5 26.7 11.5 22.0 18.3 15.9 6.6

19.6 17.9 17.3 7.8

AUC Vintra

23.1 30.7

61.3 57.6 94.8 33.6 35.0 34.2 50.5 36.1 26.0 24.1 38.2 17.7 47.5 137.2

67.9 75.2 19.2 17.7 70.1 25.3 42.7 37.8 35.1 29.8

73.0 50.9 50.9 49.7 73.1 59.2 66.3 70.7 19.9

41.5 155.1 82.9 23.1

AUC Vinter

3A4 3A4 and 2D6

3A4

3A4 3A4

3A4

0.0444

3.85

49

3.8

310

114

3A4 3A4

124

7.21

98

106

Km (μM)

3A4 3A4

3A4 and 2D6 3A4 and 2D6

2D6

3A4

3A4

2D6

CYP isoform

12

12

16

15

16

16

16

13

16

16

ref



D

a

2.5 mg 17.5 mg 100 mg 5 mg 100 mg

100 100 500 150

FCT FCT CAP UCT FCT

CAP FCT FCT FCT

mg mg mg mg

2 mg 15 mg 25 mg 50 mg 75 mg 5 mg

50 mg 10 mg 20 mg 100 mg 10 mg

CAP ODT ODT CAP FCT

CAP FCT FCT FCT CAP FCT

15 mg 30 mg 112.5 mg 15 mg 20 mg 100 mg 5 mg 125 mg 80 mg 100 mg 35 mg 1 mg 75 mg

strength

UCT UCT CAP UCT UCT FCT UCT UCT FCT FCT UCT FCT FCT

dosage forma

0.28 0.18 0.43 0.59 0.88 0.31 0.43

−0.482 −0.41 −2.608 0.19

0.2

−2.622 0.49 −1.42 0.486

1.02 1.03

−0.202 1.09

4.06 0.69

−0.44 1.22 0.24 2.02

0.59 0.35

−0.482 −0.611

−0.53 1.42

1.62 3.83 8.55 5.26 3.83 0.16 4.49 0.43

0.13 10.29

2.96


1.965 4.481 5.96 4.5 0.47 −5.642 1.28 0.486

5.065 4.86

3.53

cLogP

0.36 0.29 1.7 0.83

> 1000 833 0.6

52.7

40 29.9

0.091 625

8.33 0.4

0.36 1.9

0.008 0.00012 0.14 0.0045 6.68 39.1 0.8 0.6

0.0003 0.00058

0.027

Solubility (mg/mL)

1.852 2.299 1.961 1.211

0.00032 0.0022 0.00067 0.000040 1.111

0.147 0.00016 0.00027 0.00053 0.013 0.0011

0.926 0.035 0.070 0.080 0.167

3.704 7.407 2500.000 172.414 229.885 83.333 277.778 5.952 118.519 0.100 0.0060 0.0083 0.833

Do

0.32 0.38 0.16 0.36

632 90.3 420 4500 0.39

1.63 12625 7575 3788 81.6 924

0.637 10.0 4.99 50.7 4.14

0.80 0.40 0.000052 0.060 0.045 0.019 0.014 1.436 0.044 38.4 26.8 538.8 0.516


UCT: uncoated tablet. FCT: film coated tablet. ODT: oral disintegrating tablet. CAP: capsule. GR: granule. DS: dry syrup.

suplatast tosilate taltirelin cefcapene pivoxil hydrochloride cefdinir cefpodoxime proxetil L-carbocisteine tosufloxacin tosilate

nizatidine olopatadine hydrochloride sodium risedronate

methotrexate milnacipran hydrochloride

fluconazole lafutidine

rebamipide simvastatin terbinafine hydrochloride zaltoprofen actarit alendronate sodium anastrozole cefcapene pivoxil hydrochloride cefdinir famotidine

pranlukast quazepam

pioglitazone hydrochloride

compound

Table 1. continued

4 4 4 4

3 3 4

3

3 3

3 3

3 3

3 3

2 2 2 2 3 3 3 3

2 2

2

BCS class

100 100 500 150

2.5 17.5 100 5 100

2 15 25 50 75 5

50 10 20 100 10

15 30 112.5 15 20 100 5 125 80 100 35 1 75

dose (mg)

13.1 13.2 11.6 24.9

43.2 27.2 26.5 34.6 18.9

8.6 5.3 3.9 4.3 4.7 4.5

16.9 9.2 17.5 3.1 9.3

15.6 15.5 33.7 12.3 20.0 16.9 38.5 21.5 14.3 7.4 42.7 4.9 33.6

AUC Vintra

27.1 16.6 18.7 30.5

68.8 67.8 41.7 88.3 29.5

13.6 15.5 14.4 14.4 15.0 16.2

27.1 20.0 20.5 10.2 23.2

27.3 21.9 53.3 37.8 39.6 41.9 70.9 48.3 26.6 13.5 55.8 18.3 37.2

AUC Vinter

3A4

3A4 3A4 and 2D6

3A4 3A4

3A4 3A4

CYP isoform

21

Km (μM)

16

ref



Article

Molecular Pharmaceutics

according to the Japanese Guideline for BE study.3 As a dose, actual dose of API used for the BE study was adopted for the calculation. Therefore, the different products containing the same API have different values of Do if the actual doses of API are different. The solubility in water of APIs was measured at Sawai Pharmaceutical Co. for all drugs except ebastine and itraconazole. The solubility of those two drugs was calculated with ADMET Predictor version 7.2 (Simulation Plus, Inc., Lancaster, CA, USA). The solubility over 1000 mg/mL was denoted as 1000 mg/mL. Intra- and Intersubject Variabilities. ANOVA CV (%) was used as an index to represent intrasubject variability (Vintra) (eq 2). In eq 2, σ is the square root of the mean square error in the ANOVA table of log-transformed values.

It is important to predict the possible extent of intrasubject variability in designing the appropriate scale of a human BE study. Yamashita and Tachiki5 have investigated risk factors to incur the bioinequivalence in BE studies using physicochemical and pharmacokinetic parameters of APIs and found that, as far as BCS class 1 and class 3 drugs are concerned, there is no risk when the value of AUC/dose is greater than 18 × 10−6 h/mL. They suggested that, for oral products of such APIs, BE can be verified with a small number of subjects (usually around 24 subjects). However, if AUC/dose is less than 18 × 10−6 h/mL, a greater number of subjects might be required, due to highly variable blood concentration profiles of APIs. It was considered that the high clearance of class 1 drugs and poor membrane permeability of class 3 drugs lowered the AUC/dose, causing high intrasubject variability in oral absorption, and therefore making it difficult to prove BE. Sakuma et al.6 have also reported that, in the case of class 1 and class 3 drugs, oral formulations from different companies showed comparable levels of intrasubject variability in a BE study. This indicated that formulation-related factors are not the main causes of variability in oral absorption of highly soluble class 1 and class 3 drugs. However, in those reports, it was also noted that no clear relationship was observed between AUC/dose and the number of subjects required to prove BE for class 2 and 4 drugs. They have stated that not only factors relating to APIs but also those relating to oral formulations contribute to variations in blood concentration profiles of poorly water-soluble drugs, since the oral absorption of such drugs is governed by the in vivo dissolution profile from the formulation. In this study, 113 human BE studies on immediate release (IR) formulations of 74 compounds conducted at Sawai Pharmaceutical Co., Ltd., were analyzed with a particular focus on intra- and intersubject variabilities, to evaluate risk factors for bioinequivalence of drug products including the poorly water-soluble drugs.

ANOVA CV (%) = 100 exp(σ 2) − 1

In this study, since the results of BE study of generic products were analyzed, ANOVA CV (%) includes the variabilities caused by the differences in the formulation. However, since all generic products depicted in Table 1 were proved as BE with the reference ones in the human BE study, factors derived from the formulation differences could be neglected and the ANOVA CV (%) was considered to be usable as an index of intrasubject variability. The relative standard deviation (%) (RSD) of a test formulation in the relevant BE study was used as an index of intersubject variability (Vinter). Parameters. Maximum absorbable dose (MAD) is a simple method to estimate the maximum amount of drugs possible to be absorbed after oral administration (eq 3).7−9 MAD = Cs × Peff × SA × SITT

EXPERIMENTAL SECTION A database of various parameters for 113 oral drug products (including tablets, oral disintegrating (OD) tablets, capsules, dry syrups, and granules) of 74 compounds used for BE studies at Sawai Pharmaceutical Co., Ltd., was created for analysis and is depicted in Table 1. All BE studies included in this report were conducted by Sawai Pharmaceutical Co., Ltd., according to the Japanese Guideline for BE study3 with healthy Japanese volunteers. The protocols of all studies including blood sampling of human subject in accordance with GCP were approved by the internal review boards of Sawai Pharmaceutical Co., Ltd. BSC Classification. The Biopharmaceutics Classification System (BCS) class of the drug was determined based on dose number (Do) and cLogP, with Do calculated using the following equation: dose/Vo Cs

(3)

where Peff is the effective human intestinal permeability, SA is the effective surface area of the small intestine, and SITT is the intestinal transit time of the drug. If the MAD/dose is greater than 1, the absorption is completed within SITT (if the dissolution rate is faster than the permeation rate). Using eq 1 and eq 3, MAD/dose can be obtained as

■

Do =

(2)

P MAD SA × SITT = eff × dose Do Vo

(4)

In eq 4, because SA, SITT, and Vo are constant, MAD/dose is proportional to Peff/Do. Here, Peff/Do was defined as a new parameter, and the value corresponding to the border of MAD/ dose = 1 was obtained by substituting 800 cm2 for SA, as reported by Yu,9 3.5 h for SITT, the reported average of small intestinal transit of a solution,10,11 and 150 mL for Vo, as recommended for human BE study in Japan.3 The Peff (×10−4 cm/s) of each drug was calculated with ADMET Predictor version 7.2. Since physicochemical properties of drugs used in this study (except for pranlukast) were within the range of those of model compounds used for building the prediction model of ADMET Predictor, the calculated Peff values are considered to be reliable. As a parameter representing a metabolic activity of CYP3A4 for each substrate drug, reported values of Km obtained in the in vitro experiment using human liver microsomes were quoted from the literature.12−16 Statistical Analysis. The pharmacokinetic parameters were calculated and statistically compared with analysis of variance (ANOVA) using the computer software BESTS (version 5.0, CAC EXICARE Corp., Tokyo, Japan). The pharmacokinetic

(1)

In eq 1, dose is the actual dose strength of API (mg), Cs is the solubility of API in water (mg/mL), and Vo is the dose volume of water taken with a drug product. Although generally a Vo of 250 mL, which is recommended in the guidance released by the US Food and Drug Administration,1 is used for the calculation of Do, in this report, 150 mL was applied as Vo E


Article

Molecular Pharmaceutics parameters were logarithmically transformed before the statistical analysis. Microsoft Excel 2010 (Microsoft Corp., Redmond, WA, USA) were used for the regression analysis.

■

RESULTS 113 formulations of generic products were plotted in Figure 1 for the dose number (Do) and cLogP of their APIs. The

Figure 2. Correlation of intrasubject variability (Vintra) and intersubject variability (Vinter) in AUC. Dotted line is a regression analysis of the 113 formulations.

depended on Peff (r = −0.775, P < 0.001), indicating that low Peff is one of the factors to cause intrasubject variability in oral absorption of drugs. It was also found that the low Peff would cause high intersubject variability of class 3 drugs, showing the significant correlation between Peff and Vinter (r = −0.715, P < 0.001) in Figure 3-1b. In Figure 3-2, effects of Do on the intra- and intersubject variabilities were investigated. Vintra, of class 2 drugs, but not of other class drugs, tended to increase with increase in Do. No such tendencies were observed for Vinter. This tendency was clearly described in Figure 3-3 using the parameter of Peff/Do. In Figure 3-3, the vertical line marks a threshold value for Peff/ Do (0.149 × 10−4 cm/s) which corresponds to MAD/dose = 1 by the calculation with 800 cm2 for SA, 3.5 h for SITT, and 150 mL for Vo. In this study, 15 products of class 2 drugs showed Peff/Do < 0.149 × 10−4 cm/s (Table 1). Among those products, Vintra of 8 products were higher than 20% and of 4 products were higher than 30%. In contrast, only 3 products showed high Vintra (>20%) among other 12 products of class 2 drugs with Peff/Do > 0.149 × 10−4 cm/s. Since oral absorption of class 2 drugs with Peff/Do < 0.149 × 10−4 cm/s (thus MAD < dose) is considered to be rate-limited by drug solubility (solubilitylimited absorption), the fluctuations in the total dissolved amount of drug in the GI tract might cause high intrasubject variability in oral absorption. In Figure 2, not only drugs in class 2 or class 3 but also some drugs in class 1 showed high intra- or intersubject variability in AUC. Since factors investigated in Figure 3 cannot explain the reason for high variability with class 1 drugs, other factors that are not directly relating to the physicochemical properties of APIs might affect their blood concentration profiles after oral administration. As one such factor, effects of drug metabolizing enzymes were investigated. In Figure 4, drugs metabolized by CYP were extracted and plotted against intra- and intersubject variability in AUC. It was clearly demonstrated that drugs showing a high Vintra are substrates of CYP3A4, while those showing a high Vinter are substrates of CYP2D6. For drugs metabolized by CYP3A4, Km values reported in the literature were analyzed for possible correlation with Vintra in Figure 5. The Km value of each drug significantly correlated with Vintra in AUC (r = −0.717, P < 0.001), suggesting that a high affinity

Figure 1. BCS classification of 113 oral drug products according to dose number (Do) and cLogP. Point of intersection of axes is Do = 1 and cLogP = 1.5.

vertical and horizontal axes show cLogP and Do, respectively, and the intercept is at Do = 1 and cLogP = 1.5. In Figure 1, 55% of drug products were classified as BCS class 1, 24% as class 2, 17% as class 3. Class 4 drug products were only 4%, suggesting the difficulty to develop an oral product with this class of APIs. In this study, since dose number (Do) was calculated using the solubility which was measured in house (at Sawai Pharmaceutical Co., Ltd.) and Vo of 150 mL, some drugs change the class from the BCS database.17 In our calculation, atorvastatin, clotiazepam, glimepiride, mosapride, nilvadipine, risperidone, tamoxifen, and ticlopidine were assigned to class 1, though they are listed as class 2 in the BCS database. Also, the products of cefcapene pivoxil and cefdinir were assigned to class 3 or class 4 depending on the dose strength in each formulation. The relationship between the index for intrasubject variability (ANOVA CV, represented as Vintra) and for intersubject variability (RSD, represented as Vinter) in AUC of all drug products was demonstrated in Figure 2. Regardless of the BCS class, no appreciable correlation was observed. In the case of class 1 and class 2 drugs, some showed higher Vintra but others showed higher Vinter, while most of the class 3 drugs were plotted above the regression line, suggesting the risk of high intrasubject variability in BE study. Interestingly, class 4 drugs, which were expected to show high variability in oral absorption, showed low variability both for intra- and intersubject in this study. To understand factors which caused intra- and intersubject variability in oral absorption, in Figure 3, relations of Peff, Do, and Peff/Do with both types of variability were investigated. In Figure 3-1, no significant tendencies were observed between Peff and Vintra as a whole, whereas Vintra of class 3 drugs strongly F


Article


Figure 3. Effect of drug permeability (Peff), Do, and Peff/Do on intra- and intersubject variability in AUC. Panels 3-1, 3-2, and 3-3 show the relation of Peff, Do, and Peff/Do to Vintra (a) or Vinter (b), respectively. In panel 3-1a, the green dotted line represents a regression line only for the class 3 drugs. In panel 3-3, the vertical dotted line marks a threshold of Peff/Do (0.149 × 10−4 cm/s).

Figure 5. Effect of Km value in CYP3A4 mediated metabolism on intrasubject variability (Vintra) in AUC. Km values were quoted from the literature (refs 12−16) and were obtained in the in vitro experiment using human liver microsomes.

Figure 4. Intrasubject variability (Vintra) and intersubject variability (Vinter) in AUC of CYP3A4 and CY2D6 substrate drugs. The dotted line indicates a regression analysis of the 113 formulations.

G


Article


variability.22 In Figure 3-3, 9 of 15 products of class 2 drugs with Peff/Do < 0.149 × 10−4 cm/s showed a high value of Vintra (>20%). Since absorption of drugs with Peff/Do < 0.149 × 10−4 cm/s is considered to be limited by the solubility, this findings would suggest that class 2 drugs with solubility-limited absorption showed a high variability in oral absorption even after being formulated for clinical use. To explain the high intra- and intersubject variability observed for class 1 drugs in Figure 2, other factors than Peff or Do should be considered which affect the blood concentration profiles of the drug after oral administration. Davit et al. reviewed 1010 BE studies of 180 drugs, and they have revealed that extensive first-pass metabolism was probably the most important factor for the highly variable drugs in their analysis.23 In this study, results in Figure 4 clearly demonstrated that Vintra was high for drugs metabolized by CYP3A4 and Vinter was high for drugs metabolized by CYP2D6. Further analysis in Figure 5 revealed the impact of drug Km values for CYP3A4 on intrasubject variability in AUC. Since the Km value of each drug for the metabolizing enzyme is easy to determine with an in vitro metabolism assay usually using microsome fraction, this result may help to predict the intrasubject variability in oral drug absorption caused by CYP3A4-mediated metabolism. However, to estimate the overall impact of CYP3A4 metabolism on variability in drug absorption and pharmacokinetic profile, effects of other parameters, an intrinsic clearance of metabolism (Vmax/Km), or a fraction metabolized by the intended enzyme (fm value) should be investigated. This issue is now under investigation in which in vivo human data of pharmacokinetics of various drugs are collected and analyzed for the extent of variability. Different types of genetic polymorphisms have been reported for CYP2D6. In the Japanese, poor metabolizers (PMs) account for no more than 1% of the population, but intermediate metabolizers (IMs), with metabolic function reduced to less than half that of normal subjects, are estimated to account for 30−40%. This high incidence of IMs in the Japanese is considered to be one cause of high intersubject variability of CYP2D6 substrate drugs influenced by the distribution of PM or IM subjects in a BE study. Polymorphisms of CYP3A4 with functionally substantial consequences have not been reported so far, although polymorphisms with minor effects on enzymatic function are known. This may explain the relatively small intersubject variability seen with CYP3A4 substrate drugs. Meanwhile, CYP3A4 is an abundant isoform accounting for 40% of liver CYP and 82% of intestinal CYP in humans24 and thus greatly affects oral bioavailability of various drugs. CYP3A4 expressed in the intestine has been reported to undergo inhibition by certain food constituents such as flavonoids and catechins,25−28 which can cause day-to-day variability in CYP3A4 activity and thus be responsible for high intrasubject variability of CYP3A4 substrate drugs in this study. Therefore, for more accurate BE validation of APIs with high CYP3A4 affinity, a protocol intervention involving diet might help to reduce day-to-day variability in CYP3A4 activity. In conclusion, by analyzing the in house data of human BE studies, the risk factors for high intrasubject variability in oral drug absorption were identified as

for CYP3A4 enzyme is another factor contributing to high intrasubject variability in human BE studies.

■

DISCUSSION In this study, 113 human BE studies on IR formulations of 74 APIs conducted at Sawai Pharmaceutical Co., Ltd., were analyzed to explore the factors affecting the intrasubject variability in oral drug absorption. Because the previous report from Yamashita and Tachiki5 has determined risk factors to incur the bioinequivalence only for BCS class 1 and class 3 drugs, this study aimed to identify the factors also for class 2 drugs, with demonstrating the different aspects between intraand intersubject variability. First, relationships between intrasubject and intersubject variability for drugs in each BCS class were evaluated (Figure 2). Although no clear correlations were observed between Vintra and Vinter as a whole, it was found that class 3 drugs tended to show a larger Vintra than Vinter. In addition, analysis of the relationship between Peff and Vintra (Figure 3-1) clearly demonstrated the impact of low permeability on Vintra for class 3 drugs. Since the absorption of class 3 drugs was rate-limited by the permeability, this result might be explained by an increased intrasubject variability in absorption rate due to poor membrane permeability of drugs, as previously described by Yamashita and Tachiki.5 Tanaka et al.18 have investigated the effect of luminal fluid volume in the GI tract on oral absorption of class 1 and class 3 drugs and reported that the impact of a change in fluid volume was more remarkable for the absorption of a class 3 drug (atenolol) than for that of a class 1 drug (metoprolol). Also, since it is possible for the change in GI transit to affect the permeability-limited absorption, the variation in GI transit might be an additional factor for intrasubject variability in the absorption of BCS class 3 drugs. The fluid volume and the transit in the GI tract are known to be affected by various factors such as disease, food intake, or the excipients of product.19−21 Therefore, a risk of high variability in oral absorption should be taken into consideration in the development of an oral product if the candidate showed low permeability to the human intestine. In general, poor water solubility is considered to be one of the factors for high variability in oral drug absorption. However, in Figure 3-2, as a whole, Do showed no clear relation with both Vintra and Vinter. Similar results were also reported by Yamashita and Tachiki in which no correlations were observed between 90% confidence interval ranges of AUC or Cmax and Do.5 As one of the reasons for these observations, appropriate formulation technologies successfully improved the dissolution of poorly soluble drugs and minimized the deviation in the dissolution rates in the GI tract. Two different rate-limiting processes were involved in the incomplete oral absorption of BCS class 2 drugs, dissolutionrate-limited absorption and solubility-limited absorption. In these two rate-limiting processes, improvement of dissolution rate is achievable by relatively simple formulation technologies such as particle-size reduction. In contrast, to improve the solubility of APIs, more complicated technology is required for inducing the supersaturation in the GI tract. Those (supersaturable formulations) are salt formation, solid dispersion, cocrystallization, or the use of amorphous forms. Since the supersaturation state is unstable and influenced by many factors such as the contents in the GI tract, pH, and volume of GI fluid and the GI transit of drugs, oral drug absorption which involves the process of supersaturation sometimes shows high

(1) low permeability to the human intestine (2) solubility-limited absorption (3) high affinity to CYP3A4 H


Article


(15) Yoon, Y. J.; Kim, K. B.; Kim, H.; et al. Characterization of benidipine and its enantiomers’ metabolism by human liver cytochrome P450 enzymes. Drug Metab. Dispos. 2007, 35 (9), 1518−1524. (16) Bu, H. Z. A literature review of enzyme kinetic parameters for CYP3A4-mediated metabolic reactions of 113 drugs in human liver microsomes: structure-kinetics relationship assessment. Curr. Drug Metab. 2006, 7 (3), 231−249. (17) Access BCS Database, Drug Delivery Foundation, http://www. ddfint.org/bcs-database/. Accessed 27 Jun 2015. (18) Tanaka, Y.; Goto, T.; Kataoka, M.; Sakuma, S.; Yamashita, S. Impact of Luminal Fluid Volume on the Drug Absorption After Oral Administration: Analysis Based on In Vivo Drug Concentration-Time Profile in the Gastrointestinal Tract. J. Pharm. Sci. 2015, 104, 3120− 3127. (19) Schiller, C.; Fröhlich, C. P.; Giessmann, T.; Siegmund, W.; Mönnikes, H.; Hosten, N.; Weitschies, W. Intestinal fluid volumes and transit of dosage forms as assessed by magnetic resonance imaging. Aliment. Pharmacol. Ther. 2005, 22 (10), 971−979. (20) Sjögren, E.; Abrahamsson, B.; Augustijns, P.; Becker, D.; Bolger, M. B.; Brewster, M.; Brouwers, J.; Flanagan, T.; Harwood, M.; Heinen, C.; Holm, R.; Juretschke, H. P.; Kubbinga, M.; Lindahl, A.; Lukacova, V.; Münster, U.; Neuhoff, S.; Nguyen, M. A.; Peer, A.; Reppas, C.; Hodjegan, A. R.; Tannergren, C.; Weitschies, W.; Wilson, C.; Zane, P.; Lennernäs, H.; Langguth, P. In vivo methods for drug absorption comparative physiologies, model selection, correlations with in vitro methods (IVIVC), and applications for formulation/API/excipient characterization including food effects. Eur. J. Pharm. Sci. 2014, 57, 99−151. (21) Chen, M. L.; Straughn, A. B.; Sadrieh, N.; Meyer, M.; Faustino, P. J.; Ciavarella, A. B.; Meibohm, B.; Yates, C. R.; Hussain, A. S. A modern view of excipient effects on bioequivalence: case study of sorbitol. Pharm. Res. 2007, 24 (1), 73−80. (22) Sugano, K. Computational oral absorption simulation of free base drugs. Int. J. Pharm. 2010, 398 (1−2), 73−82. (23) Davit, B. M.; Conner, D. P.; Fabian-Fritsch, B.; et al. Highly variable drugs: observations from bioequivalence data submitted to the FDA for new generic drug applications. AAPS J. 2008, 10 (1), 148− 156. (24) Paine, M. F.; Hart, H. L.; Ludington, S. S.; Haining, R. L.; Rettie, A. E.; Zeldin, D. C. The human intestinal cytochrome P450 ″pie″. Drug Metab. Dispos. 2006, 34 (5), 880−886. (25) Harris, R. Z.; Jang, G. R.; Tsunoda, S. Dietary effects on drug metabolism and transport. Clin. Pharmacokinet. 2003, 42 (13), 1071− 1088. (26) Ho, P. C.; Saville, D. J.; Wanwimolruk, S. Inhibition of human CYP3A4 activity by grapefruit flavonoids, furanocoumarins and related compounds. J. Pharm. Pharm. Sci. 2001, 4 (3), 217−227. (27) Kimura, Y.; Ito, H.; Ohnishi, R.; Hatano, T. Inhibitory effects of polyphenols on human cytochrome P450 3A4 and 2C9 activity. Food Chem. Toxicol. 2010, 48 (1), 429−435. (28) Misaka, S.; Kawabe, K.; Onoue, S.; Werba, J. P.; Giroli, M.; Tamaki, S.; Kan, T.; Kimura, J.; Watanabe, H.; Yamada, S. Effects of green tea catechins on cytochrome P450 2B6, 2C8, 2C19, 2D6 and 3A activities in human liver and intestinal microsomes. Drug Metab. Pharmacokinet. 2013, 28 (3), 244−249.

Since these factors can be predicted from parameters obtainable by in vitro experiments, if the intended APIs have such properties, the human BE study should be designed with the risk of high intrasubject variability, thus the difficulty for proving BE, in mind.

■

AUTHOR INFORMATION

Corresponding Author

*Tel: +81 6 6105 5732. Fax: +81 6 6394 7318. E-mail: m. [email protected]. Notes

The authors declare no competing financial interest.

■

REFERENCES

(1) Guidance for Industry, Bioavailability and Bioequivalence Studies Submitted in NDAs or INDsGeneral Considerations; US Department of Health and Human Services Food and Drug Administration Center for Drug Evaluation and Research: 2014. http://www.fda.gov/ downloads/Drugs/GuidanceComplianceRegulatoryInformation/ Guidances/UCM389370.pdf. Accessed 27 Jun 2015. (2) Guideline, the investigation of bioequivalence. Committee for Medicinal Products for Human Use (CHMP); European Medicines Agency (EMA): 2010. http://www.ema.europa.eu/docs/en_GB/ document_library/Scientific_guideline/2010/01/WC500070039.pdf. Accessed 27 Jun 2015. (3) National Institute of Health Sciences Guideline, bioequivalence studies of generic products; Japanese Pharmaceutical and Food Safety Bureau, Minister of Health, Labor and Welfare: 2012. http://www.nihs.go.jp/ drug/be-guide(e)/Generic/GL-E_120229_BE.pdf. Accessed 27 Jun 2015. (4) Davit, B.; Braddy, A. C.; Conner, D. P.; Yu, L. X. International guidelines for bioequivalence of systemically available orally administered generic drug products: a survey of similarities and differences. AAPS J. 2013, 15 (4), 974−990. (5) Yamashita, S.; Tachiki, H. Analysis of risk factors in human bioequivalence study that incur bioinequivalence of oral drug products. Mol. Pharmaceutics 2009, 6 (1), 48−59. (6) Sakuma, S.; Tachiki, H.; Uchiyama, H.; et al. A perspective for biowaivers of human bioequivalence studies on the basis of the combination of the ratio of AUC to the dose and the biopharmaceutics classification system. Mol. Pharmaceutics 2011, 8 (4), 1113−1119. (7) Johnson, K. C.; Swindell, A. C. Guidance in the setting of drug particle size specifications to minimize variability in absorption. Pharm. Res. 1996, 13 (12), 1795−1798. (8) Curatolo, W. Physical chemical properties of oral drug candidates in the discovery and exploratory development settings. Pharm. Sci. Technol. Today 1998, 1 (9), 387−393. (9) Yu, L. X. An integrated model for determining causes of poor oral drug absorption. Pharm. Res. 1999, 16 (12), 1883−1887. (10) Dressman, J. B.; Amidon, G. L.; Reppas, C.; Shah, V. P. Dissolution testing as a prognostic tool for oral drug absorption: immediate release dosage forms. Pharm. Res. 1998, 15 (1), 11−22. (11) Hirtz, J. The gastrointestinal absorption of drugs in man: a review of current concepts and methods of investigation. Br. J. Clin. Pharmacol. 1985, 19 (S2), 77S−83S. (12) Yamazaki, H.; Niwa, T.; Murayama, N.; Emoto, C. Comparison of kinetic parameters for drug oxidation rates and substrate inhibition potential mediated by cytochrome P450 3A4 and 3A5. Curr. Drug Metab. 2008, 9 (1), 20−33. (13) Niwa, T.; Shiraga, T.; Ishii, I.; Kagayama, A.; Takagi, A. Contribution of human hepatic cytochrome p450 isoforms to the metabolism of psychotropic drugs. Biol. Pharm. Bull. 2005, 28 (9), 1711−1716. (14) Senda, C.; Kishimoto, W.; Sakai, K.; Nagakura, A.; Igarashi, T. Identification of human cytochrome P450 isoforms involved in the metabolism of brotizolam. Xenobiotica 1997, 27 (9), 913−922. I


Analysis of Intra- and Intersubject Variability in Oral ... - ACS Publications

Recommend Documents