The Relationship between Bitter Taste Sensor Response and Physicochemical Properties of 47 Pediatric Medicines and Their Biopharmaceutics Classification

Abstract

The purpose of this study was to investigate the relationship between response to the bitterness taste sensor and physicochemical parameters of 47 pediatric medicines and to classify these medicines according to the biopharmaceutics classification system (BCS). Forty-seven bitter compounds, most of which were on the WHO model list of essential medicines for children (March 2017), were used in the study. Solutions (0.1 mM) were evaluated by an artificial taste sensor using membranes sensitive to bitterness. On the basis of principal component analysis of taste sensor measurements, chlorpromazine, haloperidol, propranolol, amitriptyline, diphenhydramine were predicted to express the strongest levels of basic bitterness, surpassing that of quinine. Correlation tests between bitter taste sensor outputs and physicochemical properties were then carried out and the compounds classified in terms of their biopharmaceutical properties. High log P values (≥2.82), physiological charge (≥1), low log S values (<−3) and small polar surface area (PSA; <45.59 Ų) were found to correlate significantly with the responses of bitter taste sensors. Forty-one of the 47 compounds could be placed into one of four groups in the BCS, on the basis of dose number (D₀), an indicator of solubility which takes into account clinical dosage, and fractional absorption (Fa). For medicines classified in group 4, the factors D₀ > 1 and Fa < 0.85 significantly correlated with the responses of the taste sensor for basic bitterness. It was concluded that lipophilicity, physiological charge, solubility, PSA and D₀ are the main factors affecting the bitterness of pediatric medicines.

Introduction

Child-appropriate drug formulations are required for the efficacious and safe therapy of diseases in childhood.¹⁾ In general, child-appropriate drug formulations have some key attributes; for example, the ability to deliver precise drug doses, acceptable palatability, tablet size, etc. New solid dosage forms with improved key attributes such as mini-tablets (defined as tablets with both diameter and height smaller than 3 mm), dispersible or orodispersible tablets, and orodispersible films, have been developed for pediatric use.^2–4) These modern formulations are often considered to be the dosage forms of choice.⁴⁾ However, solid dosage forms may not be suitable for drug administration to neonates⁵⁾; thus, oral liquid dosage forms may be the most convenient dosage form for the treatment of pediatric patients, especially neonates. Solubility, taste, stability and bioavailability are the major issues associated with the suitability of oral liquid dosage forms for pediatric use.⁶⁾

Avoidance or masking of bitterness is one of the biggest challenges in the development of pharmaceuticals for children, who are often unable to tolerate bitter-tasting drugs. This thwarts drug delivery and may, in some cases, be life-threatning.^7,8) Taste-masking is a challenging task. Different approaches have been used to overcome this challenge, including spheronization, encapsulation, ion exchange, granulation, and the use of taste suppressant, sweeteners and flavors.^9–14)

Among these techniques, the simplest and most frequently used approach is the addition of flavors and sweeteners. However, in practice, bitterness suppression of active ingredients in pharmaceuticals is far from simple. Humans have as many as 25 TASTE2 receptors (hTAS2R), a family of G protein-coupled receptors which detect bitterness.¹⁵⁾ Some bitter compounds are selective for a single hTAS2R, while others activate multiple hTAS2Rs.¹⁶⁾ This fact means the diversity of bitterness.

The taste sensor with lipid/polymer membranes developed by Toko is widely used in pharmaceutical applications.^17–21) Bitter materials are absorbed on the hydrophobic part of the membrane and cause a change in membrane potential caused by changing the charge density. Lipid/polymer membrane selectivity for each taste was improved by modulating both the hydrophobic interaction between the taste sensor and bitter substance and the membrane charge density. There are several types of bitterness or astringency sensors which have different components of lipids and plasticizers: BT0 for hydrochloride salts including quinine hydrochloride¹⁸⁾; AN0, AC0, for basic materials, such as amlodipine besilate¹⁹⁾ or solifenacin succinate²⁰⁾; C00 for acidic bitter materials such as diclofenac sodium and etodolac which belong to the non-steroidal anti-inflammatory drug (NSAIDs)²¹⁾; AE1 for astringent materials such as tannic acid. Correlation between results from the taste sensor and taste intensities evaluated by human sensation testing has proven the usefulness of the taste sensor in predicting the taste of various substances. These taste sensors are now widely applied for development of taste masking formulation. Pres et al. reported that taste masked formulation of cetirizine hydrochloride has been achieved using β-cyclodextrin and cherry/sucralose flavor system.²²⁾ Taste masking effect of carrageenan on donepezil hydrochloride was also revealed by using taste sensor and it was achieved to develop the commercial product.²³⁾ Taste sensing system data can be used to provide guidance on the selection of taste-masking formulations.

However, it is also true that some bitter substances would be detected by none of taste sensors now available. It is observed as a matter of course because of the wide range (25 kinds) of human taste receptors as mentioned above. New type of taste sensor which can detect those bitterness is still under development. Classifying the bitter substances by those properties would be helpful to achieve the development.

Accordingly, the purpose of this study was to evaluate the bitterness of some important pharmaceuticals in pediatric medicine and to identify the factors associated with their bitterness in terms of physicochemical properties and biopharmaceutical class. The 45 oral formulations selected from the WHO model list of essential medicines for children (6th edition, 2017) and 2 oral formulations; chlorpheniramine and diphenhydramine broadly used for children as anti-allergic medicine in Japan, total 47 oral formulations were investigated in this study.

Experimental

Materials

The following 47 compounds were purchased from Wako Pure Chemical Industries, Ltd., Osaka, Japan for use in this study: acetylsalicylic acid, acyclovir, allopurinol, amitriptyline, amoxicillin, azathioprine, azithromycin, budesonide, caffeine, carbamazepine, chloramphenicol, chlorpheniramine maleate salt, chlorpromazine, diethylcarbamazine, digoxin, diphenhydramine hydrochloride, doxycycline, erythromycin, ethosuximide, fluconazole, folic acid, furosemide, haloperidol, hydrochlorothiazide, hydrocortisone, ibuprofen, isoniazid, loratadine, metformin, metoclopramide, metronidazole, nitrofurantoin, ondansetron, oseltamivir, paracetamol, phenytoin, potassium iodide, prednisolone, primaquine, propranolol, propylthiouracil, quinine hydrochloride, salbutamol, spironolactone, sulfamethoxazole, valproic acid and vancomycin.

Methods

Taste Sensor Measurement

The taste sensor SA402B (Intelligent Sensor Technology Inc., Atsugi, Japan) was used to evaluate drug bitterness.^17–21) The lipid/polymer membranes used as bitter taste sensors in this study were as follows: for detecting basic bitterness, AC0 (hexadecanoic acid/dioctyl phenylphosphonate), AN0 (phosphoric acid di-n-decyl ester/dioctyl phenylphosphonate), BT0 (phosphoric acid di-n-decyl ester/bis(1-butylpentyl) adipate, tributyl O-acetylcitrate); for detecting astringency, AE1 (tetradodecylammonium bromide/dioctyl phenylphosphonate); and for detecting acidic bitterness, C00 (tetradodecylammonium bromide/2-nitrophenyl octyl ether).¹⁷⁾

The 47 compounds studied were each dissolved in water to form solutions of 0.1 mM for testing in the bitter taste sensor. The taste sensor outputs of the 47 compounds were analyzed by principal component analysis (PCA) using Ekuseru–Toukei 2010 (Social Survey Research Information Co., Ltd.).

Physicochemical Parameters

The following physicochemical parameters of the 47 compounds were derived from the DRUGBANK database (https://www.drugbank.ca/): molecular weight (MW), water solubility (WS), log P, log S (predicted by ALOGPS), physiological charge, polar surface area (PSA), refractivity and polarizability.

Biopharmaceutic Classification

The biopharmaceutics classification system (BCS) is a scientific framework for classifying drug substances based on their aqueous solubility and intestinal permeability.^24,25) With respect to solubility, the dose number (D₀) is defined as^{24, 26, 27)}:

  

Drugs with D₀ ≤ 1 and D₀ > 1 are considered to have high and low solubility, respectively. With respect to permeability, a drug substance is considered to be highly permeable when its systemic bioavailability or extent of absorption in humans is ≥85% of an administered dose, based on mass balance or in comparison to an intravenous reference dose.²⁸⁾

In this study, D₀ was calculated using the highest strength dose referenced in the WHO model list of essential medicines for children and the solubility was extracted from the DRUGBANK database, as was fractional absorption (Fa) in humans as an indicator for permeability. Group 1 drugs have both high solubility and permeability (D₀ ≤ 1, Fa ≥ 0.85); Group 2 drugs have low solubility and high permeability (D₀ > 1, Fa ≥ 0.85); Group 3 drugs have high solubility and low permeability (D₀ ≤ 1, Fa ≤ 0.85); Group 4 drugs have both low solubility and permeability (D₀ > 1, Fa ≤ 0.85).

Correlation between Physicochemical Parameters and Taste Sensor Responses

In determination of correlation coefficients, the physicochemical parameters were replaced by either 1 (high) or 0 (low) as follows according to the method described in a previous article²⁹⁾: If the parameter of a compound was equal to or larger than that of quinine, the parameter was given the value 1; if the parameter was less than that of quinine, it was given the value 0. Taste sensor (AN0, AC0, BT0, C00, AE1) outputs for each of the 47 compounds were replaced by either 1 (response) or 0 (no response) as follows: if the taste sensor response value at 0.1 mM was ≥2 mV, the response of the compound was designated 1. If the response value at 0.1 mM was <2 mV, the response of the compound was designated 0. Correlation coefficients between physicochemical parameters and taste sensor responses were calculated and Fisher’s exact tests carried out using Ekuseru–Toukei, 2010.

Correlation between Biopharmaceutics Classification and Taste Sensor Responses

Fisher’s exact tests between drugs classified in group 1, 2, 3 or 4 according to the BCS and taste sensor responses (1 or 0) as determined above were carried out using Ekuseru–Toukei, 2010.

Results and Discussion

Taste of 47 Compounds Predicted by Taste Sensor

The taste map obtained by PCA based on the five taste sensor outputs is shown in Fig. 1. PCA was used to estimate the largest and second-largest relative contributing factors (PC1 and PC2, respectively). The relative contributions of PC1 and PC2 were 52.1% and 35.3%, respectively. Taste sensor outputs of AN0, AC0 and BT0 (for basic bitterness) contributed to PC1 rather than PC2, while those of AE1 and C00 (for astringency or acidic bitterness, respectively) contributed to PC2 rather than PC1. Those mean that the direction of the PC1 axis reflects the contribution of basic bitterness and the direction of the PC2 axis reflects the contribution of acidic bitterness or astringency.

Fig. 1. PCA of Taste Sensor Outputs (AC0, AN0, BT0, C00, AE1) in Response to 47 Compounds (0.1 mM)

(For abbreviations, see Table 1.)

Chlorpromazine (No. 13), haloperidol (No. 23), propranolol (No. 40), amitriptyline (No. 4) and diphenhydramine (No. 16) were predicted to express strong basic bitterness, surpassing that of quinine (No. 42), which was used as the standard of bitterness. Ondansetron (No. 33), primaquine (No. 39), doxycycline (No. 17) and metoclopramide (No. 30) were also predicted to be basic bitter compounds.

Potassium iodide (No. 37), furosemide (No. 22) and azithromycin (No. 7) responded to both AE1 and C00. Chlorpheniramine (No. 12) responded to AE1 and BT0 but not to C00. It was predicted that potassium iodide, furosemide and azithromycin would show astringency and acidic bitterness, while chlorpheniramine would show astringency and basic bitterness.

Physicochemical Parameters

The following physicochemical properties of the 47 compounds were extracted from the DRUGBANK database: MW, log P, log S (predicted by ALOGPS), physiological charge, PSA, refractivity and polarizability. These are summarized in Table 2.

Table 1. Pediatric Pharmaceutical Medicines Used in the Study

1. Acetylsalicylic acid (ASA)

2. Acyclovir (ACV)

3. Allopurinol (ALP)

4. Amitriptyline (AMT)

5. Amoxicillin (AMX)

6. Azathioprine (AZP)

7. Azithromycin (AZM)

8. Budesonide (BUD)

9. Caffeine (CAF)

10. Carbamazepine (CBZ)

11. Chloramphenicol (CPH)

12. Chlorpheniramine maleate salt (CPA)

13. Chlorpromazine (CPZ)

14. Diethylcarbamazine (DCZ)

15. Digoxin (DGX)

16. Diphenhydramine hydrochloride (DPH)

17. Doxycycline (DXC)

18. Erythromycin (ETM)

19. Ethosuximide (ESX)

20. Fluconazole (FCZ)

21. Folic acid (FOL)

22. Furosemide (FUS)

23. Haloperidol (HPD)

24. Hydrochlorothiazide (HCT)

25. Hydrocortisone (HCS)

26. Ibuprofen (IBU)

27. Isoniazid (ISZ)

28. Loratadine (LRD)

29. Metformin (MEF)

30. Metoclopramide (MCP)

31. Metronidazole (MNZ)

32. Nitrofurantoin (NFT)

33. Ondansetron (ODS)

34. Oseltamivir (OTV)

35. Paracetamol (PCM)

36. Phenytoin (PHN)

37. Potassium iodide (PTS)

38. Prednisolone (PSL)

39. Primaquine (PRQ)

40. Propranolol (PPL)

41. Propylthiouracil (PTU)

42. Quinine hydrochloride (QUI)

43. Salbutamol (SBM)

44. Spironolactone (SPL)

45. Sulfamethoxazole (SMX)

46. Valproic acid (VPA)

47. Vancomycin (VCM)

Table 2. Physicochemical Properties of 47 Pharmaceutical Medicines (Data Extracted from DRUGBANK)

Compounds	MW	log P	log S	Charge	PSA	Refractivity	Polarizability
1 Acetylsalicylic acid	180	1.43	−2.1	−1	63.6	44.45	17.1
2 Acyclovir	225	−0.95	−1.4	0	114.76	54.63	21.51
3 Allopurinol	136	−1.7	−1.4	0	65.85	54.24	11.67
4 Amitriptyline	277	5.1	−4.8	1	3.24	101.51	33.74
5 Amoxicillin	365	0.75	−2.6	0	132.96	89.5	35.53
6 Azathioprine	277	0.84	−2.4	0	118.1	70.95	24.26
7 Azithromycin	749	3.03	−3.2	2	180.08	194.11	83.11
8 Budesonide	430	2.42	−4	0	93.06	116.11	47.11
9 Caffeine	194	−0.24	−1.2	0	58.44	49.83	18.95
10 Carbamazepine	236	2.1	−3.2	0	46.33	71.89	25
11 Chloramphenicol	323	1.15	−2.8	0	115.38	73.2	28.08
12 Chlorpheniramine	275	3.74	−3.7	1	16.13	80.85	30.82
13 Chlorpromazine	319	5.18	−4.9	1	6.48	93.76	35.1
14 Diethylcarbamazine	199	0.9	0.07	0	26.79	58.28	22.89
15 Digoxin	780	1.04	−3.8	0	203.06	193.23	84.8
16 Diphenhydramine	255	3.44	−3.5	1	12.47	79.93	29.86
17 Doxycycline	444	−0.72	−2.8	0	181.62	113.89	43.65
18 Erythromycin	734	2.37	−3.2	1	193.91	186.04	78.21
19 Ethosuximide	141	0.1	−0.15	0	46.17	35.96	14.45
20 Fluconazole	306	0.58	−2.3	0	81.65	97.2	26.52
21 Folic acid	441	−0.04	−3.8	−2	208.99	111.01	42.06
22 Furosemide	330	2.71	−3.4	−1	122.63	77.47	30.55
23 Haloperidol	375	3.7	−4.9	1	40.54	102.59	39.15
24 Hydrochlorothiazide	297	−0.16	−2.1	0	118.36	63.11	25.35
25 Hydrocortisone	362	1.79	−3.3	0	94.83	97.4	39.45
26 Ibuprofen	206	3.5	−3.5	−1	37.3	60.73	23.76
27 Isoniazid	137	−0.71	−0.59	0	68.01	37.46	13.21
28 Loratadine	382	4.8	−4.5	0	42.43	116.98	41.67
29 Metformin	129	−1.8	−2	2	88.99	56.64	13.43
30 Metoclopramide	299	2.18	−3	1	67.59	83.52	32.7
31 Metronidazole	171	−0.15	−1.5	0	83.87	41.22	15.82
32 Nitrofurantoin	238	0.03	−2.8	0	118.05	52.11	20.49
33 Ondansetron	293	2.56	−3.1	1	39.82	86.78	33.16
34 Oseltamivir	312	1.3	−2.7	1	90.65	84.2	34.65
35 Paracetamol	151	0.51	−1.6	0	49.33	42.9	15.52
36 Phenytoin	252	2.26	−3.6	0	58.2	70.18	25.48
37 Potassium iodide	166	—	—	1	0	0	1.78
38 Prednisolone	360	1.66	−3.2	0	94.83	98.49	38.69
39 Primaquine	259	2.76	−3.7	1	60.17	78.51	29.92
40 Propranolol	259	3.03	−3.5	1	41.49	76.83	29.98
41 Propylthiouracil	170	1.53	−2.6	0	41.13	48.9	17.79
42 Quinine	324	2.82	−3	1	45.59	94.69	35.96
43 Salbutamol	239	0.44	−2	1	72.72	67.87	26.86
44 Spironolactone	416	3.1	−5.3	0	60.44	113.5	46.03
45 Sulfamethoxazole	253	0.79	−2.7	−1	98.22	64.5	24.99
46 Valproic acid	144	2.54	−1.8	−1	37.3	40.25	17
47 Vancomycin	1449	1.11	−3.8	1	530.49	346.61	138.7

The MW values of amoxicillin, azithromycin, budesonide, digoxin, doxycycline, erythromycin, folic acid, furosemide, haloperidol, hydrocortisone, loratadine, prednisolone, spironolactone and vancomycin are larger than that of quinine (MW 324). The log P values of amitriptyline, azithromycin, chlorpheniramine, chlorpromazine, diphenhydramine, haloperidol, ibuprofen, loratadine, primaquine, propranolol and spironolactone are greater than that of quinine (log P 2.82). The log S values of acetylsalicylic acid, acyclovir, allopurinol, amoxicillin, azathioprine, caffeine, chloramphenicol, diethylcarbamazine, doxycycline, ethosuximide, fluconazole, hydrochlorothiazide, isoniazid, metformin, metoclopramide, metronidazole, nitrofurantoin, oseltamivir, paracetamol, propylthiouracil, salbutamol, sulfamethoxazole and valproic acid are greater than that of quinine (log S −3). The physiological charge of amitriptyline, azithromycin, chlorpheniramine, chlorpromazine, diphenhydramine, erythromycin, haloperidol, ibuprofen, metformin, metoclopramide, ondansetron, oseltamivir, potassium iodide, primaquine, propranolol, salbutamol and vancomycin are greater than that of quinine hydrochloride (1). The PSA of amitriptyline, chlorpheniramine, chlorpromazine, diethylcarbamazine, diphenhydramine, haloperidol, ibuprofen, loratadine, ondansetron, potassium iodide, propranolol, propylthiouracil and valproic acid are smaller than that of quinine (PSA 45.59 Ų). The refractivity of amitriptyline, azithromycin, budesonide, digoxin, doxycycline, erythromycin, fluconazole, folic acid, haloperidol, hydrocortisone, loratadine, prednisolone, spironolactone and vancomycin are greater than that of quinine (94.69 m³·mol⁻¹). The polarizability of azithromycin, budesonide, digoxin, doxycycline, erythromycin, folic acid, haloperidol, hydrocortisone, loratadine, prednisolone, spironolactone and vancomycin are greater than that of quinine (35.96 ų).

Fisher’s extract tests were carried out to evaluate the correlations between the physicochemical parameters and taste sensor responses (Table 3). High log P values (≥2.82) and physiological charge (≥1) significantly correlated with the responses of AN0, AC0 and BT0, respectively (p < 0.01). Low log S values (<−3) significantly correlated with the responses of BT0 (p < 0.01). Small PSA (< 45.59 Ų) significantly correlated with the responses of AC0 and BT0 (p < 0.01). There were no significant correlations between refractivity or polarizability and the responses of the taste sensor.

Table 3. Correlation Coefficients between Physicochemical Parameters and Taste Sensor Responses

	MW≥324	log P≥2.82	Log S < −3	Charge ≥ 1	PSA < 45.59 Å2	Refractivity ≥ 94.69 m3·mol−1	Polarizability ≥ 35.96 Å3
AC0	−0.03	0.68a)	0.26	0.54a)	0.55a)	0.09	0.03
AN0	0.09	0.42a)	0.14	0.54a)	0.29	0.22	0.16
BT0	0.13	0.70a)	0.47a)	0.78a)	0.57a)	0.24	0.21
C00	−0.10	−0.25	−0.22	−0.03	−0.25	−0.10	−0.13
AE1	−0.12	−0.06	−0.03	0.11	−0.06	−0.25	−0.20

a) p < 0.01, Fisher′s exact test.

Judging from the fact that high log P values (≥2.82) and physiological charge (≥1), low log S values (<−3) and small PSA (<45.59 Ų) significantly correlated with the responses of basic bitter taste sensors, it can be concluded that high lipophilicity, high physiological charge, low solubility and small PSA are factors associated with basic bitterness (Tables 2, 3).

Bitter compounds are absorbed on the hydrophobic part of the taste sensor membrane and cause a change in membrane potential by changing the charge density.¹⁷⁾ Roland et al. suggested that hydrophobic interaction is important for binding to bitter taste receptor hTAS2R14 which is broadly tuned by a large number of agonists.^30,31) These references support our finding of a relation between high lipophilicity (low solubility) or physiological charge and bitterness.

PSA, defined as the sum of surfaces of polar atoms in a molecule, is known as a parameter predictive of intestinal absorption.³²⁾ Our results suggested that low PSA value (<45.59 Ų) would be a factor linked with bitterness, although there have been no previous reports about a relationship between PSA and bitterness. Clark³³⁾ suggested that a poorly absorbed molecule will have PSA >140 Ų. In our study, only five of the 47 compounds had a PSA >140 Ų, with four of the five also having poor absorption (Fa < 0.85).

Molar refractivity is related to dispersive interaction which is one of the most important forces (hydrophobic, dispersive and electrostatic interaction) in biochemical ligand binding, and the molecular orbital charge distribution or the electrostatic potential at the van der Waals radius may be used for modeling the electrostatic interaction.³⁴⁾ There was no significant correlation between refractivity and bitterness in our study, while there was significant correlation between log P, which is related to hydrophobic interaction, and bitterness. From these results, it was concluded that hydrophobic interaction has a larger factor loading in relation to bitterness than dispersive or electrostatic interaction.

AC0, AN0 and BT0 all of these are composed of negatively charged membrane but there is difference in lipids or plasticizers were developed to detect the basic bitterness which have high hydrophobicity. The difference of the correlations between physicochemical parameters and taste sensor responses among three taste sensors for basic bitterness suggests that selection of taste sensor suitable for each bitter substance would improve the accuracy of prediction for bitterness.

Biopharmaceutics Classification

Given the differences in the administered dosage for the pediatric medicines, the biopharmaceutics classification, which uses D₀ and Fa, was used in addition to physicochemical properties, to evaluate correlations with bitterness. The highest strength dose (mg), water solubility (mg/mL), D₀ and Fa values of the 47 compounds are summarized in Table 4. The highest strength doses of each compound were obtained from the WHO model list of essential medicines for children (6th list, 2017) or medical package insert, while water solubilities and Fa values were extracted from the DRUGBANK database. D₀ values of each compound except potassium iodide (No. 37) were calculated from the highest strength dose and water solubility. No water solubility data are given for potassium iodide in the DRUGBANK database, and no Fa values are given for folic acid (No. 21), nitrofurantoin (No. 32), potassium iodide (No. 37), primaquine (No. 39), salbutamol (No. 43) and valproic acid (No. 46) (see Table 4). A map of biopharmaceutics classifications of the 41 compounds with Fa values, constructed from D₀ and Fa, is shown in Fig. 2(a) and the drug classifications are listed in Fig. 2(b).

Table 4. Parameters for Biopharmaceutics Classification

Compounds	Highest strength dose (mg)	Water solubility (mg/mL)	Dose number (D0)	Fractional absorption (Fa)
1 Acetylsalicylic acid	500	1.460	1.370	1
2 Acyclovir	200	9.080	0.088	0.1–0.2
3 Allopurinol	300	5.880	0.204	0.9
4 Amitriptyline	25	0.005	22.222	0.3–0.6
5 Amoxicillin	250	0.958	1.044	1
6 Azathioprine	50	1.070	0.187	1
7 Azithromycin	500	0.514	3.891	0.37
8 Budesonide	0.2	0.046	0.018	1
9 Caffeine	10	11.000	0.004	1
10 Carbamazepine	100	0.152	2.632	0.89
11 Chloramphenicol	250	0.461	2.169	0.8
12 Chlorpheniramine	8	0.052	0.617	1
13 Chlorpromazine	25	0.004	23.981	1
14 Diethylcarbamazine	100	236.000	0.002	1
15 Digoxin	0.25	0.127	0.008	0.7–0.85
16 Diphenhydramine	150	0.075	7.979	0.4–0.6
17 Doxycycline	100	0.630	0.635	1
18 Erythromycin	300	0.459	2.614	0.18–0.45
19 Ethosuximide	250	101.000	0.010	0.93
20 Fluconazole	50	1.390	0.144	0.9
21 Folic acid	5	0.076	0.263	—
22 Furosemide	20	0.118	0.678	0.6
23 Haloperidol	2	0.004	1.794	0.4–0.75
24 Hydrochlorothiazide	25	2.240	0.045	0.5–0.6
25 Hydrocortisone	20	0.199	0.402	0.95
26 Ibuprofen	600	0.068	35.088	1
27 Isoniazid	300	34.900	0.034	1
28 Loratadine	10	0.013	2.985	1
29 Metformin	1000	1.380	2.899	0.5–0.6
30 Metoclopramide	10	0.310	0.129	0.8
31 Metronidazole	500	5.920	0.338	0.8
32 Nitrofurantoin	100	0.415	0.964	—
33 Ondansetron	8	0.248	0.129	0.56–0.6
34 Oseltamivir	75	0.686	0.437	0.75
35 Paracetamol	500	4.150	0.482	0.88
36 Phenytoin	100	0.071	5.626	0.7–1
37 Potassium iodide	60	—	—	—
38 Prednisolone	25	0.239	0.418	1
39 Primaquine	15	0.056	1.064	—
40 Propranolol	40	0.079	2.015	1
41 Propylthiouracil	50	0.466	0.429	1
42 Quinine	300	0.334	3.593	0.76–0.88
43 Salbutamol	5	2.150	0.009	—
44 Spironolactone	25	0.002	50.505	1
45 Sulfamethoxazole	400	0.459	3.486	1
46 Valproic acid	500	2.360	0.847	—
47 Vancomycin	500	0.225	8.889	0.6

Fig. 2. (a) Biopharmaceutics Classification Map Constructed Using Values of D₀ and Fa

●: Compounds predicted to have greater basic bitterness than quinine. (b) Compounds classified in groups. Group 1 (D₀ ≤ 1 and Fa > 0.85), Group 2 (D₀ > 1 and Fa > 0.85), Group 3 (D₀ ≤ 1 and Fa < 0.85) and Group 4 (D₀ > 1 and Fa < 0.85).

Fisher’s extract tests were carried out to evaluate the correlations between biopharmaceutics classification and taste sensor response (Table 5). Classification in group 4 significantly correlated with response to BT0 (p < 0.01) or AN0 (p < 0.05). Incidentally, there were significant correlations between D₀ > 1 and response to BT0 (p < 0.01) or AN0 (p < 0.05), and between Fa < 0.85 and response to BT0 (p < 0.05). These results suggest that BT0 and AN0 can detect the bitterness of substances with low solubility (D₀ > 1) and low permeability (Fa < 0.85).

Table 5. Correlation Coefficients between Biopharmaceutics Classification and Taste Sensor Responses

	Group 1	Group 2	Group 3	Group 4
AC0	−0.18	0.03	−0.11	0.29
AN0	−0.18	−0.10	−0.11	0.43b)
BT0	−0.37	−0.04	−0.01	0.49a)
C00	0.19	−0.25	0.05	0.00
AE1	−0.02	−0.06	0.22	−0.12

a) p < 0.01, b) p < 0.05, Fisher’s exact test.

Our results, showing that both D₀ > 1 (compounds in group 2 or 4) and Fa < 0.85 (compounds in group 3 or 4) are related to a basic bitter taste sensor output, lead to the largest correlation being found between classification in group 4 and response to basic bitter taste sensor. The statistical significance of the relationship between D₀ ≤ 1 and response to taste sensor BT0 (p = 0.0073) or AN0 (p = 0.0376), was greater than that between Fa < 0.85 and response to taste sensor BT0 (p = 0.0152) or AN0 (p = 0.1302). It was therefore considered that solubility (D₀) has a larger factor loading for bitterness than permeability (Fa). In practice, all the drugs (quinine, chlorpromazine, haloperidol, propranolol, amitriptyline and diphenhydramine) which were predicted to have a strong bitter taste (shown in Fig. 2 (a) as black points) had D₀ > 1.

Conclusion

The purpose of this study was to investigate the main factors determining bitterness in 47 pediatric medicines on the basis of their physicochemical properties and to classify these medicines using the BCS. Bitterness was predicted by taste sensor SA402B which is broadly used for pharmaceutical formulations and confirmed the accuracy of prediction in previous reports by human sensation tests.^19,21) Therefore, our predictions would be reliable although human sensation tests were not carried out in this study. The physicochemical properties, high lipophilicity, high physiological charge, low solubility and small PSA were found to be the factors with the strongest association with basic bitterness. Thus, the most bitter medicines were classified in group 4 in the BCS (D₀ > 1 and Fa < 0.85), with D₀, as an indicator of solubility, having a larger factor loading for bitterness than permeability (Fa). Chlorpromazine, haloperidol, propranolol, amitriptyline and diphenhydramine were predicted to express strong basic bitterness, surpassing that of quinine, and indeed, all the compounds predicted to express strong bitterness, including quinine, had D₀ > 1. These findings may be helpful not only for the classification of bitterness and the search for suitable bitterness-suppressing strategies, but also for predicting the bitterness of novel drugs.

Acknowledgments

This work was supported by JSPS KAKENHI Grant Number 16K08426.

Conflict of Interest

The authors declare no conflict of interest.

References

• 1) Thabet Y., Klingmann V., Breitkreutz J., J. Clin. Pharmacol., 58 (Suppl 10), S26–S35 (2018).
• 2) Van Riet-Nales D. A., Kozarewicz P., Aylward B., de Vries R., Egberts T. C., Rademaker C. M., Schobben A. F., AAPS PharmSciTech, 18, 241–249 (2017).
• 3) Slavkova M., Breitkreutz J., Eur. J. Pharm. Sci., 75, 2–9 (2015).
• 4) Preis M., AAPS PharmSciTech, 16, 234–241 (2015).
• 5) Batchelor H. K., Fotaki N., Klein S., Adv. Drug Deliv. Rev., 73, 102–126 (2014).
• 6) Amin F., Khan S., Shah S. M. H., Rahim H., Hussain Z., Sohail M., Ullah R., Alsaid M. S., Shahat A. A., Drug Des. Devel. Ther., 12, 3855–3866 (2018).
• 7) Mennella J. A., Beauchamp G. K., Clin. Ther., 30, 2120–2132 (2008).
• 8) Kendrick J. G., Ma K., Dezorzi P., Hamilton D., Can. J. Hosp. Pharm., 65, 196–201 (2012).
• 9) Petrovick G. F., Breitkreutz J., Pein-Hackelbusch M., Int. J. Pharm., 506, 361–370 (2016).
• 10) Han C. S., Kim S., Oh D. W., Yoon J. Y., Park E. S., Rhee Y. S., Kim J. Y., Shin D. H., Kim D. W., Park C. W., Materials, 12, E1000 (2019).
• 11) Tan D. C. T., Ong J. J., Gokhale R., Heng P. W. S., Int. J. Pharm., 547, 385–394 (2018).
• 12) Sharma S., Lewis S. Int. J. Pharm. Pharm. Sci., 2, 6–13 (2010).
• 13) Abraham J., Mathew F., Int. J. Pharm., 6, 12–19 (2014).
• 14) Faisal W., Farag F., Abdellatif A. A. H., Abbas A., Curr. Drug Deliv., 15, 167–185 (2018).
• 15) Meyerhof W., Batram C., Kuhn C., Brockhoff A., Chudoba E., Bufe B., Appendino G., Behrens M., Chem. Senses, 35, 157–170 (2010).
• 16) Di Pizio A., Niv M. Y., Bioorg. Med. Chem., 23, 4082–4091 (2015).
• 17) Kobayashi Y., Habara M., Ikezaki H., Chen R., Naito Y., Toko K., Sensors, 10, 3411–3443 (2010).
• 18) Ito M., Yoshida M., Kobayashi Y., Hiraoka M., Ikezaki H., Uchida T., Sensors and Materials, 23, 483–492 (2011).
• 19) Uchida T., Yoshida M., Hazekawa M., Haraguchi T., Furuno H., Teraoka M., Ikezaki H., J. Pharm. Pharmacol., 65, 1312–1320 (2013).
• 20) Yoshida M., Haraguchi T., Uchida T., Chem. Pharm. Bull., 62, 1252–1258 (2014).
• 21) Haraguchi T., Miyazaki A., Yoshida M., Uchida T., J. Pharm. Pharmacol., 65, 980–987 (2013).
• 22) Preis M., Grother L., Axe P., Breitkreutz J., Int. J. Pharm., 491, 8–16 (2015).
• 23) Harada T., Sakurai M., Hondo S., Yasui M., Owaki T., Yakugaku Zasshi, 134, 325–331 (2014).
• 24) Amidon G. L., Lennernas H., Shah V. P., Crison J. R., Pharm. Res., 12, 413–420 (1995).
• 25) Abdel-Rahman S. M., Amidon G. L., Kaul A., Lukacova V., Vinks A. A., Knipp G. T., Clin. Ther., 34, S11–S24 (2012).
• 26) Benet L. Z., Broccatelli F., Oprea T. I., AAPS J., 13, 519–547 (2011).
• 27) Dave R. A., Morris M. E., Int. J. Pharm., 511, 111–126 (2016).
• 28) FDA, “Waiver of in vivo bioavailability and bioequivalence studies for immediate-release solid oral dosage forms based on a biopharmaceutics classification system guidance for industry.” 2017.
• 29) Haraguchi T., Uchida T., Yoshida M., Kojima H., Habara M., Ikezaki H., Chem. Pharm. Bull., 66, 71–77 (2018).
• 30) Roland W. S., van Buren L., Gruppen H., Driesse M., Gouka R. J., Smit G., Vincken J. P., J. Agric. Food Chem., 61, 10454–10466 (2013).
• 31) Born S., Levit A., Niv M. Y., Meyerhof W., Behrens M., J. Neurosci., 33, 201–213 (2013).
• 32) Ertl P., Rohde B., Selzer P., J. Med. Chem., 43, 3714–3717 (2000).
• 33) Clark D. E., J. Pharm. Sci., 88, 807–814 (1999).
• 34) Ghose A. K., Pritchett A., Crippen G. M., J. Comput. Chem., 9, 80–90 (1988).