See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/261765581

The Maillard reaction of Bisoprolol fumarate with various reducing

carbohydrates

Article  in  European journal of pharmaceutical sciences: official journal of the European Federation for Pharmaceutical Sciences · April 2014
DOI: 10.1016/j.ejps.2014.04.005 · Source: PubMed

CITATIONS READS

9 2,297

5 authors, including:

Mateusz Szalka Jacek Lubczak

Bausch Health Companies Inc. Rzeszów University of Technology
3 PUBLICATIONS   19 CITATIONS    107 PUBLICATIONS   935 CITATIONS   

SEE PROFILE SEE PROFILE

Dorota Narog Marek Laskowski

Rzeszów University of Technology Wyższa Szkoła Informatyki i Zarządzania z siedzibą w Rzeszowie
13 PUBLICATIONS   132 CITATIONS    2 PUBLICATIONS   11 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Oxygen activation by transition metal complexes View project

The Maillard reaction of Bisoprolol fumarate with various reducing carbohydrates View project

All content following this page was uploaded by Jacek Lubczak on 06 November 2017.

The user has requested enhancement of the downloaded file.

 European Journal of Pharmaceutical Sciences 59 (2014) 1–11

Contents lists available at ScienceDirect

European Journal of Pharmaceutical Sciences

journal homepage: www.elsevier.com/locate/ejps

The Maillard reaction of bisoprolol fumarate with various

reducing carbohydrates
Mateusz Szalka a,⇑, Jacek Lubczak b, Dorota Naróg c, Marek Laskowski a, Krzysztof Kaczmarski d
a
Product Development Department, ICN Polfa Rzeszów S.A., 35-959 Rzeszów, Poland
b
Department of Organic Chemistry, Rzeszów University of Technology, 35-959 Rzeszów, Poland
c
Department of Physical Chemistry, Rzeszów University of Technology, 35-959 Rzeszów, Poland
d
Department of Chemical and Process Engineering, Rzeszów University of Technology, 35-959 Rzeszów, Poland

a r t i c l e i n f o a b s t r a c t

Article history: HPLC analysis of drug products containing bisoprolol fumarate and lactose revealed the presence of N-
Received 25 September 2013 formylbisoprolol, which is a final product of the Maillard reaction. Formulations containing secondary
Received in revised form 31 March 2014 amines and reducing carbohydrates are prone to the condensation of amine and carbonyl functional
Accepted 7 April 2014
groups and formation of glycosylamines in pharmaceutically relevant conditions. Further rearrangement
Available online 16 April 2014
occurs in the presence of a nucleophile and leads to the formation of 1-deoxy-1-amino-2-ketose also
known as the Amadori Rearrangement Product (ARP). The influence of water content, carbohydrate,
Keywords:
and lubricant types on the reaction rate was tested. The reaction progress was monitored by HPLC and
Maillard reaction
Amadori rearrangement
UV–Vis spectrophotometry. The structures of intermediates were confirmed by the LC/MS2 analysis.
Bisoprolol fumarate N-formylbisoprolol – the final reaction product – was synthesised and characterised by LC/MS2, H1 and
Glycosylamine C13 NMR.
1-Deoxy-1-amino-2-ketose Ó 2014 Elsevier B.V. All rights reserved.
Excipient incompatibility

1. Introduction
Common incompatibilities between the active pharmaceutical
ingredients and excipients (tablet diluents) occur due to the
Maillard reaction also known as a non-enzymatic browning reac-
tion. Over 100 years ago Louis Maillard reported, that carbonyl
compounds may react with amino acids and proteins to form the
complex mixtures of brown pigments with a characteristic malt-
like odour (Maillard, 1912). This reaction has been studied for
many years and is described in details in the food and nutritional
literature (Ledl and Schleicher, 1990; Ledl, 1990). Nowadays, the
Maillard reaction was also extensively studied in the pharmaceuti-
cal systems (Qiu et al., 2005; Monajjemzadeh et al., 2009) where
glucose, lactose, and maltose are commonly used as fillers in
Abbreviations: ACE, angiotensin converting enzyme inhibitors; Al/PVdC, alu-
minium/polyvinylidene chloride; Ar, aromatic ring; ARP, Amadori Rearrangement
Product; DMSO, dimethylsulphoxide; EMEA, European Medicines Agency; FDA,
Food and Drug Administration; GC, gas chromatography; HPLC, high performance
liquid chromatography; ICH, The International Conference on Harmonisation of
Technical Requirements; IR, infrared spectroscopy; KF, Karl Fischer titration; MS,
mass spectrometry; NMR, Nuclear Magnetic Resonance; RT, retention time; UV–Vis,
ultraviolet–visible spectrophotometry.
⇑ Corresponding author. Tel.: +48 795 521 562.
E-mail address: mateusz.szalka@valeant.com (M. Szalka).
capsules and tablets (Rowe et al., 2003). In the drug product formu-
lations the non-enzymatic browning reaction occurs between a
reducing carbohydrate (mainly mono- or disaccharide) which
may exist in open-chain aldehyde forms, and active substances
containing primary or secondary amine functional groups (Hodge
and Rist, 1953). For many years tertiary amines were not consid-
ered as substrates for the Maillard reaction. However, as described
by Thumma and Repka (2009), reaction of promethazine and lac-
tose also results in non-enzymatic browning.
Bisoprolol fumarate belongs to the beta-blocker class of thera-
peutic substances and is commonly used in a treatment of hyper-
tension, coronary failure, and arrhythmia (Merck KGaA Darmstadt
Germany, 2001). Chemically bisoprolol is a derivative of phenoxya-
minopropanol (secondary amine); it is used as a racemic mixture
for therapeutic purposes. The ConcorÒ family of drug products
was originally developed by the Merck Group and is currently reg-
istered in over 90 countries worldwide. Recently, there are many
generic formulations of bisoprolol fumarate available in the
European market, most of them contain reducing carbohydrates
as primary tablet fillers – mainly lactose due to its low price and
good compressibility.
The European Society of Cardiology strengthened the position of
beta blockers in a treatment of the chronic and acute hart failures
in the revised guidelines (McMurray et al., 2012). There is evidence

http://dx.doi.org/10.1016/j.ejps.2014.04.005
0928-0987/Ó 2014 Elsevier B.V. All rights reserved.
2 M. Szalka et al. / European Journal of Pharmaceutical Sciences 59 (2014) 1–11
that beta-blockers are complementary with diuretics or ACE inhib-
itors (the list of abbreviations is given in the first page of this
paper), and may be used simultaneously in a treatment of the diag-
nosed heart failure. Thus, we may expect that many new generic
formulations containing two active substances from the different
classes will be developed and registered at relevant authorities.
The formulation scientists should not forget that a combination
of amine and carbonyl compounds may result in the non-enzy-
matic browning reaction and formation of undesirable chemical
compounds which often demonstrate a biological activity.
The first reaction step involves condensation of a carbonyl
functional group of carbohydrate and an amine group from active
substance to form a glycosylamine and water. The second step of
this reaction was proposed by Amadori, who proved that heating
D-glucose with primary amines leads to the formation of two iso-
mers: one labile which readily undergoes rearrangement to the
another more stable isomer (Amadori, 1925). Both structures of
the labile and stable isomers were later identified as a glycosyl-
amine and 1-deoxy-1-amino-2-ketose, respectively (Kuhn and
Dansi, 1936; Kuhn and Weygand, 1937). Further stages of the
non-enzymatic browning may occur through many different reac-
tion pathways, including the fragmentation and dehydration of
carbohydrates or the cyclisation and formation of aromatic com-
pounds (Ledl et al., 1986). The non-enzymatic browning was
observed at slow rates in the neutral and acidic conditions. How-
ever, as reported by Hodge and Rist (1953), alkaline conditions
are the most favourable for this reaction. Studies of the Maillard
reaction in a liquid model system confirm, that the addition of
small amounts of hydroxides significantly improve the reaction
rate and may lead to the formation of new degradation products.
For example the formation of N-formylfluoxetine was observed
in the reaction of lactose and fluoxetine hydrochloride only after
the addition of potassium hydroxide (Wirth et al., 1998). It is
important to highlight, that glycosylamines and Amadori Rear-
rangement Products (ARP) are susceptible to the hydrolysis. Thus,
if the kinetic study of the Maillard reaction is being performed by
tracing the ARP, then the reaction conditions should exclude aque-
ous solutions (Wolfrom et al., 1949; Mitts and Hixon, 1944).
The general mechanism of the Maillard reaction (glycosylation)
for glucose and a secondary amine is presented in Fig. 1a. The
mechanism of the Amadori rearrangement in a presence of the
catalytic amounts of acid was originally proposed by Kuhn and
Weygand (Kuhn and Dansi, 1936) and involved a protonation of
nitrogen in the glycosylamine, a formation of ammonium ion
which stays in the equilibrium with a Schiff base and then further
rearrangement to 1-deoxy-1-amino-2-ketose. This mechanism was
also presented by Hodge (1955), who remarked that the attraction
of a proton by nitrogen is more probable than by a weakly basic
oxygen. In 1950s Gottschalk (1952) and Isbell and Frush (1958)
proposed the mechanism of the Amadori rearrangement which
Fig. 1. Generally proposed: (a) The Maillard reaction pathway – glycosylation and (b) mechanism of the Amadori rearrangement in acidic conditions.
included the protonation of oxygen atom, immediate ring opening
of glycosylamine and further rearrangement to the Amadori prod-
uct (ARP). Simon and Kraus (1970) in their mechanistic studies on
the Amadori rearrangement point that O-protonation is much
more favourable for the ring opening than N-protonation. In fact,
protonation of the oxygen atom turns it into a better leaving group,
thus, the ring opening should occur more readily. It is difficult to
unambiguously determine according to which mechanism the
Amadori rearrangement occurs in pharmaceutical systems. It
may also be possible that final reaction mechanism includes at first
attraction of a proton by the basic nitrogen, and when transfer of a
proton to the oxygen atom and further rearrangement. The general
mechanism of the Amadori rearrangement proposed by Kuhn and
Weygand is presented in Fig. 1b.
The scope of our study is to demonstrate the Maillard/Amadori
reactions of bisoprolol with various reducing carbohydrates in the
solid model systems. Which were designed to test the influence of
water, various tablet diluents, and lubricants on the reaction rate.
We are also alerting that the Maillard/Amadori reactions may
occur in the commercially available tablets stored in the climatic
chambers with accelerated conditions as indicated by the Interna-
tional Conference on Harmonisation (ICH) (European Medicines
Agency, 2003).
It is difficult to carry out the complete kinetic studies because of
the complexity of both the Maillard reaction and the Amadori rear-
rangement. In addition, the intermediates and final products of
conversion may degrade according to several reaction pathways,
which often lead to the regeneration of amine moiety (Labuza
and Massaro, 1990; Labuza and Baisier, 1992). All obtained results
were carefully considered and used in a preparation of the theoret-
ical reaction mechanism.
In our studies the reaction was monitored by the reversed phase
HPLC and UV–Vis spectrophotometry. Water content was
determined by the Karl Fischer titration. The molecular masses of
intermediates were confirmed by LC/MS2. The structure of the final
product was compared to a synthesised working standard and con-
firmed by the LC/MS2 and NMR analysis.
2. Materials and methods
2.1. Materials
Bisoprolol fumarate was obtained from ICN Polfa Rzeszów S.A.,
Poland.
Potassium dihydrogen phosphate (anhydrous), formic acid, and
sodium hydroxide were obtained from POCH, Poland.
Acetonitrile HPLC grade, dimethylsulphoxide for GC, isopropyl
alcohol HPLC grade, Karl Fischer titration reagent 5.0, Karl Fischer
methanol-free solvent were obtained from Merck GmbH Germany.
 M. Szalka et al. / European Journal of Pharmaceutical Sciences 59 (2014) 1–11
Triethylamine >99.0% was obtained from Sigma Aldrich, UK.
All excipients used for the preparation of solid mixtures and
tablets were commercially available. The excipients were previ-
ously tested according to the appropriate Ph. Eur. monographs
and comply with all requirements.
Glucose monohydrate and maize starch were obtained from
Roquette, France.
Crystalline maltose advantose 100 was obtained from SPI
Pharma, France.
Lactose anhydrous was obtained from Friesland Campino DMV
BV, Netherlands.
Lactose monohydrate Super Tab 11SD was obtained from DMV
Fonterra Excipients GmbH & Co., Germany.
Cellulose microcrystalline VIVAPUR type 200 and sodium stea-
ryl fumarate PRUVÒ were obtained from J.R.S. Pharma, Germany.
Calcium phosphate was obtained from cfb Budenheim,
Germany.
Magnesium stearate was obtained from Faci S.p.a., Italy.
Sorbitol Parteck SI 400, stearic acid, talc were obtained from
Merck, Germany.
Silica colloidal hydrophobic Aerosil R972 Pharma was obtained
from Evonik Degussa GmbH, Germany.
Two popular drug products (brand A and B) containing bisopro-
lol fumarate (2.5 mg strength) and lactose were obtained from the
European market, both products were unexpired.
The Synergi Hydro-RP column 250  4.6 mm was obtained from
Phenomenex, Shim-Pol, Poland.
Karl Fischer titration was performed on the Mettler Toledo Vol-
umetric KF Titrator, the procedure according to the Ph. Eur. general
method. Approximately 0.5 g of a solid mixture was used for the
determination of water content.
UV–Vis analysis was performed on the Unicam 500 UV–Visible
spectrophotometer, wavelength 490 nm, cuvette with optical path
1 cm, test solutions contained equivalence of 1.0 mg/ml of bisopro-
lol fumarate.
2.2. HPLC–UV method
The high performance liquid chromatography analysis was
performed on a Waters 2695 Separations Module (max operating
pressure 350 bar) with autosampler (25 ll syringe size) and the
photodiode array detector (2998) from Waters Corporation
(Milford, MA). The column Synergi Hydro-RP 250  4.6 mm with
pre-column was maintained at 50 °C, the autosampler temperature
was set at 25 °C, UV detector wavelength 225 nm. Two eluents
were used: the 0.05 M phosphate buffer pH 3.1 (A) and acetonitrile
HPLC grade (B). The flow rate was set to 1.0 ml/min. The gradient
pump mode with initial conditions 75:25 A:B for 10 min, linear
change to 20:80 A:B at 23 min, then returned to 75:25 A:B at
24 min and held until the equilibration of initial conditions. The
analysis run time was 30 min, the injection volume was 10 ll.
Approximate retention times: 7.5 min Amadori Rearrangement
Product (ARP), 7.8 min N-substituted glycosylamine (Maillard
Reaction Product), 11.8 min Bisoprolol, 21.3 min N-formylbisopro-
lol. Obtained results were evaluated by the Empower 3 software.
HPLC/MS analysis was performed on the 1200 Infinity Series
Agilent Technologies 6460 Triple Quadrupole LC/MS–MS system:
electron-spray ionisation mode, positive ionisation, capillary volt-
age 4000 V, collision energy 30 eV, fragmentor 10 (scan mode),
fragmentor 135 (product ion mode). Above described HPLC condi-
tions were applied, except that phosphate buffer was not compat-
ible with MS detector, thus, it was substituted by a mixture of
water and formic acid (pH 3.1).
Nuclear Magnetic Resonance (NMR) analysis was performed on
the FT-NMR Bruker Avance 500 MHz, Ultra Shield 500 MHz with
3
hexamethyldisiloxane as an internal standard. D2O was added in
some experiments to identify the interchangeable protons.
2.3. The preparation of bisoprolol base
Bisoprolol fumarate (25 g) was dissolved in 125 ml of redistilled
water; sodium hydroxide 30% aqueous solution (13 ml) was added
dropwise and mixed. This cloudy solution was extracted twice
with methylene chloride (60 ml and 40 ml). The organic layers
were combined and washed twice with redistilled water
(2  40 ml), the organic layers were combined again and dried
with anhydrous magnesium sulphate for 30 min. Mixture was
filtered to remove the drying agent; methylene chloride was
evaporated using a rotary evaporator to obtain the oil residue. Vac-
uum pump was used to remove the residual methylene chloride.
Calculated molecular weight 325.4 was confirmed by LC/MS
precursor ion mass (m/z) [M+H]+ = 326.1 (formation of potassium
adduct was observed [M+K]+ = 364.0).
IR (KBr plates) main signals: 3380 cm 1 (strong broad signal
due to OAH and weak sharp signal due to NAH); 3030 cm 1
(weak signal ArAH bonding); 2855–2960 cm 1 (strong alkane
CAH signals); 1879, 1743, 1709, 1607 (oscillation aromatic ring),
1580 (oscillation aromatic ring), 1506 (oscillation aromatic ring),
1456 (CH2 deformation vibrations), 1375 (CH3 deformation
vibrations), 1328 (˜CHA), 1295 (@CAOAC), 1250 (CAOAH), 1171
(tertiary CAOH), 1089 (non-planar in para-substituted phenylene
group),1031 (@CAOAC), 842 and 786 cm 1 (non-planar in para-
substituted phenylene group).
1
H NMR (500 MHz, CDCl3) 7.24–7.27 (d, 2H), 6.86–6.89 (d, 2H),
4.50 (s, 2H), 4.09–4.14 (m, 1H), 3.93–3.99 (m, 2H), 3.58–3.59
(broad, 2H, interchangeable protons: AOH and @NH), 3.59–3.63
(m, 1H), 3.57 (s, 4H), 2.84–2.91 (m, 2H), 2.72–2.76 (m, 1H),
1.16–1.17 (d, 6H), 1.11–1.12 (d, 6H).
13
C NMR (125 MHz, CDCl3) d = 158.22, 130.84, 129.32, 114.40,
72.81, 71.87, 70.57, 69.45, 68.04, 67.45, 53.45, 49.27, 22.48,
22.43, 22.08.
2.4. The preparation of N-formylbisoprolol
Bisoprolol base (23.09 g) was dissolved in 115 ml of dimethyl-
formamide, a clear solution was obtained and heated under reflux
for 48 h (approximate temperature 150 °C). The dark reaction mix-
ture was cooled to the ambient temperature and filtered. The
excess of a solvent was removed from the filtrate with vacuum
pump. N-formylbisoprolol was separated from the reaction mix-
ture by methylene chloride – water extraction. The organic layers
were combined, dried with anhydrous magnesium sulphate
(2.0 g) for 2 h and filtered; methylene chloride was evaporated
using a rotary evaporator. A vacuum pump was used to remove
the residual solvent. The purity tested by HPLC method was
99.1%. Calculated molecular weight 353.4 was confirmed by LC/
MS precursor ion mass (m/z) [M+H]+ = 354.1, LC/MS fragmentation
pattern confirmed the formylation of amine functional group (m/
z): 250.1; 144.1; 102.0.
IR (KBr plates) main signals: 3390 cm 1 (strong broad signal
due to OAH); 3030 cm 1 (weak signal ArAH bonding); 2855–
2960 cm 1 (strong alkane CAH signals); 2620 cm 1 (weak signal
CAH aldehyde); 1882, 1766, 1649 (strong signal carbonyl group,
C@O), 1606 & 1580 & 1506 (oscillation aromatic ring), 1456 (CH2
deformation vibrations) , 1417 (aldehyde group), 1365 (CH3 defor-
mation vibrations), 1327 (˜CHA), 1293 (@CAOAC), 1243
(CAOAH), 1180 (tertiary CAOH), 1089 (non-planar in para-substi-
tuted phenylene group), 1036 (@CAOAC), 843 and 776 cm 1
(non-planar in para-substituted phenylene group).
1
H NMR (500 MHz, CDCl3) 8.24 (s, 1H), 7.26–7.28 (d, 2H), 6.86–
6.89 (d, 2H), 4.50 (s, 2H), 4.23 (broad s, 1H), 4.09–4.16 (m, 1H),
4 M. Szalka et al. / European Journal of Pharmaceutical Sciences 59 (2014) 1–11
4.01–4.08 (m, 1H), 3.80–3.90 (m, 2H), 3.58–3.67 (m, 1H), 3.60 (s,
4H), 3.47–3.52 (m, 2H), 1.23–1.30 (dd, 6H), 1.16–1.18 (d, 6H).
13
C NMR (125 MHz, CDCl3) d = 164.80, 157.89, 131.04, 129.37,
114.36, 72.78, 71.88, 71.03, 69.49, 69.36, 67.45, 51.15, 45.92,
22.08, 21.86, 21.71.
2.5. Studies of the Maillard reaction and Amadori rearrangement in a
liquid state
Sodium hydroxide (assay P98.8%, amounts 0, 50, 100, 150,
200 mg) was transferred to a 50 ml reaction vessel and combined
with 20 ml of propan-2-ol; the initial temperature was set at
75 °C. The mixture was vigorously stirred until complete dissolu-
tion of sodium hydroxide (final solutions with concentration 0,
0.025, 0.05, 0.075, or 0.1 M of NaOH were obtained). Dimethyl-
sulphoxide (30 ml) was added to the reaction vessel and the tem-
perature was increased to the final reaction temperature (90, 100,
or 110 °C). Glucose (5.58, 8.37, or 11.16 g) was transferred to the
reaction vessel; immediately after complete dissolution of a carbo-
hydrate the bisoprolol base (0.5 g) was added. The reaction was
carried out for 5 h, after every 30 min 1.0 ml of a reaction mixture
was withdrawn from the reaction vessel and diluted to 10 ml with
the mobile phase. Working standards of the glycosylamine and
Amadori Rearrangement Product were not available, thus, their
concentration was calculated in relation to bisoprolol fumarate
standard. The concentration of N-formylbisoprolol was calculated
in relation to the synthesised working standard.
The reaction vessel was jacketed cylindrical vessel with approx-
imate volume of 50 ml, connected to the thermostated bath; glyc-
erol was used as a circulating heating medium to ensure high
reaction temperatures. The temperature was controlled with a
qualified thermometer. Vigorous stirring of a reaction mixture
was obtained by applying a magnetic stirrer. The reaction vessel
was tightly closed with a glass stopper to prevent evaporation of
reaction solvents. To simplify the reaction model we used bisopro-
lol base (free amine) instead of bisoprolol fumarate.
2.6. Screening studies
Solid samples containing 5% (w/w) of the active substance were
prepared by adding 7.5 g of bisoprolol fumarate to 141 g of a
carbohydrate (glucose, lactose anhydrous, lactose monohydrate,
maltose monohydrate) and mixed. In the last step 1.5 g of magne-
sium stearate was added and mixed shortly. About 1.0 g of a solid
mixture was transferred to 20 ml Headspace vial and closed tightly
with a septum. The samples were placed in a well equilibrated
oven at 95 °C and removed after 1, 2, 3, 4, and 6 h of heating. The
test solutions were prepared by weighing 200 mg of a solid mix-
ture (equivalence of 10 mg of bisoprolol fumarate) into 10.0 ml
volumetric flask, diluted with HPLC mobile phase and sonicated
for 15 min. Samples were then centrifuged at 4000 rpm for
10 min, the supernatant was transferred to a glass vial and
analysed by HPLC. Since there are no working standards of the
glycosylamine or Amadori Rearrangement Product available, we
have assumed that both compounds have the same UV-response
as bisoprolol. The browning observed in each solution was also
analysed with the UV–Vis spectrophotometer.
2.7. The Maillard reaction/Amadori rearrangement in tablets –
pharmaceutical development
This study was designed to test the influence of various types of
tablet fillers and lubricants, which are commonly used as excipi-
ents in the pharmaceutical industry. Also the influence of ratio of
active substance to tablet filler was tested. For this purpose the
solid mixtures containing 5%, 20%, and 40% (w/w) of bisoprolol
fumarate and tablet filler were prepared. Appropriate amounts of
the active substance and a tested excipient were weighed to the
container, closed with a lid and mixed for 60 min; in the final stage
1% of a lubricant was added and mixed shortly. The wide range for
bisoprolol content was applied to avoid ambiguous results, how-
ever, reference and generic drug products contain approximately
5% of the active substance. Tablets were prepared by compressing
the solid mixtures with Korsch PH 106 Rotary Tablet Press using
6 mm stamps. The average hardness of tablets was 100 N.
The tablets and loose solid mixtures were transferred into the
polyethylene containers closed with a lid and placed in the moni-
tored climatic chambers at 30 °C/65%RH, 40 °C/75%RH, and in a
laboratory oven at 50 °C. No desiccators were used. The relative
humidity in a laboratory oven measured manually was approx.
10%RH. All samples were tested by HPLC at the beginning of stabil-
ity studies and after 3 and 6 months of storage.
Lubricants tested: colloidal silica, talc, sodium stearyl fumarate,
magnesium stearate, stearic acid. Above listed lubricants are com-
monly used to enable tableting of powders, by preventing sticking
of tablets to the stamps. Although these compounds have no
reducing properties, some of them may catalyse the Maillard
reaction.
Tablet fillers tested: cellulose microcrystalline, maize starch,
lactose anhydrous, lactose monohydrate, maltose monohydrate,
glucose monohydrate, sorbitol, and calcium phosphate. From
above mentioned fillers only lactose, maltose, and glucose may
exist in open-chain forms with free aldehyde group, thus, all three
are reducing sugars. Cellulose and maize starch are long polymers
and contain very few terminal carbonyl groups, both compounds
are not reducing carbohydrates. Sorbitol is a sugar alcohol, it has
no reducing properties. However, sorbitol is obtained by the reduc-
tion of aldehyde group of glucose into a hydroxyl group, small
amounts of unreacted glucose may be present in the marketed
product. Calcium phosphate (CaHPO4) is an inorganic compound
and also has no reducing properties.
3. Results and discussion
3.1. Reactions in a liquid state
The reaction of glucose monohydrate and bisoprolol in a liquid
state was initially carried out in the aqueous-ethanolic solution
heated at 60 °C as described by Wirth (Wirth et al., 1998). The
browning of solutions was observed indicating that the Maillard/
Amadori rearrangement occurs in these conditions. However, the
studies in a mixture of water and ethanol have two significant
disadvantages. First, that a reaction temperature should be main-
tained at relatively low level, below boiling point of water–ethanol
solution, whereas the rate of the Maillard reaction strongly
depends on the temperature. The second important disadvantage
is that both the glycosylamine and 1-deoxy-1-amino-2-ketose
are very susceptible to hydrolysis. In fact, after the addition of
potassium hydroxide formation of N-formylbisoprolol was
observed, however after 30 min the reaction has reached its equi-
librium and did not proceed further. N-formylated bisoprolol is
also susceptible to hydrolysis in the alkaline conditions leading
to the bisoprolol base and formic acid, which slowly neutralizes
reaction conditions. It was assumed that removal of water from
the reaction mixture would improve the Maillard/Amadori rear-
rangement, however most organic solvents were restricted due
to the low solubility of carbohydrates and hydroxides. In higher
temperatures sufficient amount of glucose may be dissolved in
dimethylsulphoxide, whereas sodium hydroxide may be dissolved
in propan-2-ol. In the further studies mixture of 30 ml DMSO and
20 ml propan-2-ol was used to assure high reaction temperature
 M. Szalka et al. / European Journal of Pharmaceutical Sciences 59 (2014) 1–11
(up to 110 °C), good solubility of substrates, and low water content.
It is difficult to obtain water free conditions as some amounts of
water may be introduced to the reaction vessel with dimethyl-
sulphoxide and glucose.
The first step of the Maillard reaction includes condensation of
the amine group of bisoprolol and the carbonyl group of glucose.
This reaction was studied at three temperatures (90, 100, 110 °C)
and with excess of glucose. The analysis by HPLC revealed the
presence of a glycosylamine (7.8 min) on a chromatogram of the
reaction mixture after 30 min of heating, molecular weight 488.1
was confirmed by the LC/MS analysis. Fig. 2a represents LC/MS
spectra of a glycosylamine.
As the reaction proceeded, new peak (7.5 min) appeared on a
chromatogram of the reaction mixture. Co-elution indicated close
structural similarity which was also confirmed by the LC/MS
studies (front of a peak was analysed). New peak was assigned to
the Amadori Rearrangement Product; its molecular weight 488
[M+H]+ (m/z) was confirmed by LC/MS studies. Tjan and
Ouweland (1974) in their investigation into the structure of ARPs
discovered that these products may exist in the equilibrium in
the pyranose and furanose forms. The detailed studies proved that
pyranose form is more favoured, probably because the six member
rings are more stable. The pyranose form exists exclusively in the b
conformation, it may be assumed that a large amine moiety will
always adopt the equatorial position due to the large 1,3-diaxial
interactions. Nevertheless, the presence of the furanose form in
both a and b conformations was also confirmed in Tjan’s studies.
Different HPLC columns were tested, however, the best separa-
tion was obtained by applying the Synergi Hydro-RP C18 with
polar endcapping, which provides the additional retention for polar
compounds. The initial HPLC method contained methanol in the
mobile phase, however, this solvent was too strong as the glycosyl-
amine and ARP eluted with front of the mobile phase. Attempts of
the HPLC analysis with acetonitrile gave better separation. The MS
fragmentation patterns obtained for both glycosylamine and ARP
peaks were very similar. Some differences in the intensity of
Fig. 2. LC/MS spectra of glycosylamine (a) and Amadori Rearrangement Product (b).
5
signals were noticed. First of all, the precursor ions of both com-
pounds have molecular weight 488 [M+H]+ (m/z) and readily
undergo two consecutive dehydrations to form ions 470 and 452
(m/z), which were found on the fragmentation spectrum. The
double dehydration was also observed by Wang et al. in their MS
studies of Amadori compounds (Wang et al., 2008). In our MS con-
ditions the cleavage of the covalent bond between phenyl oxygen
and neighbouring carbon atom led to the formation of fragment
ions with molecular weight 278 (m/z) which may also undergo
two dehydrations to ions 260 and 242 (m/z). The LC/MS spectra
of glycosylamine and ARP are presented in Fig. 2a and b, respec-
tively. Formation of potassium adducts was observed during LC/
MS analysis.
In the liquid model system without sodium hydroxide the con-
centration of glycosylamine and ARP remained low even after 5 h
of heating. N-formylbisoprolol was not observed in these
conditions. However, the addition of NaOH significantly improved
reaction rates and formation of N-formylbisoprolol. This fact may
lead to a conclusion that nucleophile plays significant role in the
Maillard/Amadori rearrangement. There are information in a liter-
ature that small amounts of water may also promote the reaction,
however, no mechanistic explanation was provided (Serajuddin
et al., 1999). Experiments in liquid model system were repeated
with the addition of 50 mg (0.025 M) of sodium hydroxide. In these
conditions the formation of the glycosylamine was much faster, as
was also expected, the rate of the Amadori rearrangement
increased. The fastest reaction was observed in 110 °C. After 5 h
of heating the concentrations of glycosylamine and ARP in all
tested temperatures (110, 100, and 90 °C) was similar indicating
that the equilibrium state was achieved. N-formylbisoprolol was
observed as the final reaction product, however, the biggest
amount of this compound was formed in 90 °C. In high tempera-
tures hydrolysis in the alkaline conditions occurs faster than the
formation of N-formylated bisoprolol. The influence of various con-
centrations of hydroxide ions on the reaction rate was also tested.
The additional experiments with 100, 150, and 200 mg of sodium
6 M. Szalka et al. / European Journal of Pharmaceutical Sciences 59 (2014) 1–11
hydroxide (0.05, 0.075, and 0.1 M solutions respectively) were car-
ried out in 90 °C. The biggest amounts of glycosylamine and ARP
were formed when small amounts of a catalyst were used
(0.025 M NaOH). Both compounds were observed at lower concen-
trations in 0.05, 0.075, or 0.1 M sodium hydroxide solutions, prob-
ably because rate of alkaline hydrolysis improved. Finally, the
smallest amount of ARP was formed in the reaction without the
catalyst. It may be concluded that the Maillard reaction and Ama-
dori rearrangement are readily catalysed by small amounts of
nucleophiles.
The main degradation pathway of ARP leads to N-formylbiso-
prolol, acknowledged as the final reaction product. Formamides
may arise either by the thermal decomposition of the Amadori
Rearrangement Products as described by Mills et al. (1970) or by
oxidative fission of C1AC2 bond in a carbohydrate moiety as
described by Weygand and Bergmann (1947). The second degrada-
tion pathway was also confirmed byKawakishi and Cheng (1994),
who studied oxidative decomposition of ARP catalysed by Copper
(II) ions. Fig. 3 presents the rates of the formation of N-formylbiso-
prolol in reaction mixtures with various concentrations of sodium
hydroxide. Noteworthy, in all experiments carried out with glucose
and sodium hydroxide additional peaks were observed on a chro-
matogram of the reaction mixture. Compounds were identified as
glycosylamines formed from C5, C4, and C3 sugar moieties. Their
molecular weights (m/z) were respectively 458.1 (RT 8.6 min),
428.1 (RT 9.4 min), and 398.1 (RT 10.7 min). The degradation of
Maillard products by the fragmentation of carbohydrate molecule
was also observed by Ledl et al. (1986).
Additionally, the influence of a carbohydrate type on the rate of
Maillard reaction and Amadori rearrangement was tested. For this
purpose 0.03 mol of glucose, lactose, or maltose were dissolved in a
reaction mixture and heated in 110 °C for 5 h. The reaction was
catalysed by 50 mg of sodium hydroxide. The concentration of gly-
cosylamine was similar for lactose and maltose, and much lower
for glucose. It was confirmed by the HPLC analysis that the Amado-
ri rearrangement proceeds much faster for disaccharides than for a
monosaccharide. Finally, the biggest amounts of N-formylbisopro-
lol were also observed for maltose and lactose, whereas only small
amount of this final product was found in the reaction mixture
with glucose. It seems that structural arrangement plays an impor-
tant role in the studied chemical pathway.
Based on the obtained results a proposition for the mechanism
of the Maillard reaction and Amadori rearrangement between
Fig. 3. The rate of the formation of N-formylbisoprolol in reaction mixtures with
various concentrations of hydroxide ions at 90 °C.
bisoprolol and glucose was presented in Fig. 4. The mechanism
represents the reaction in solid mixtures (tablets, capsules), where
slightly acidic conditions may be found.
The condensation of bisoprolol and glucose leads to the forma-
tion of a glycosylamine and water, this step is promoted by the
high temperature. There is no prove weather N-protonation or
O-protonation occurs first in the acidic conditions, both mecha-
nism are reasonable. However, results obtained in the liquid state
confirm that the presence of a nucleophile is important in the Mail-
lard reaction and Amadori rearrangement. In alkaline conditions
protonation of glycosylamine cannot occur, nevertheless, hydrox-
ide ions are strong enough nucleophiles to attract a proton from
C2 position and induce rearrangement. Similarly in reactions in
solid state water attracts a proton from C2 position and forces
the Amadori rearrangement which further occurs spontaneously
via keto-enol tautomerism to form ARP. Water molecule is about
16 k times less reactive nucleophile than hydroxide ion, thus N-
or O-protonation may be necessary to improve reaction rate, this
explains why reaction in solid state is readily catalysed by acids.
The ARP does not exist as an open chain structure; it quickly
undergoes rearrangement to the pyranose or furanose forms which
coexist in equilibrium. The mechanism of the Amadori rearrange-
ment is very similar to the mechanism of the formation of enamine
from a ketone and a secondary amine, except that the presence of
hydroxyl group next to double bond forces keto-enol tautomerism.
3.2. Screening studies
The formulation scientists should perform simple screening
tests during a pharmaceutical development to detect interactions
between active substances and excipients. The heating of the solid
mixture for few hours is usually sufficient to observe intensive
browning and malt-like odour – the characteristic signs of the
Maillard reaction. This method was widely used in the past to
determine formulation incompatibilities (Castello and Mattock,
1962; Hammouda and Salakawy, 1971). The browning may be
additionally measured by UV spectrophotometry at 490 nm wave-
length (Martins and Van Boekel, 2003), this approach was used in
our studies with the HPLC analysis to detect at which stage of the
Maillard/Amadori rearrangement brown colour appears. The solid
mixtures containing 5% (w/w) of bisoprolol fumarate and appropri-
ate carbohydrate were heated in 95 °C for 6 h in a tightly closed
headspace vials to protect water from evaporation. The set of
experiments for lactose anhydrous, lactose monohydrate, maltose
monohydrate, and glucose monohydrate was carried out. In the
assumed time points vials were removed from a thermostated
oven. Test solutions containing equivalence of 1.0 mg/ml of biso-
prolol fumarate were prepared as described in the experimental
section and analysed by HPLC and UV spectrophotometry (mobile
phase was used as UV reference solution).
Results for monosaccharide: The most intensive degradation of
bisoprolol was observed in the samples with glucose, it may be
easily explained because molar ratio of glucose to API is twice
bigger than the molar ratio of lactose or maltose to the active sub-
stance. The glycosylamine undergoes the Amadori rearrangement,
however, further formation of N-formylbisoprolol proceeds very
slowly. In fact, the smallest amount of the final reaction product
was observed when glucose was used as filler. Solid mixtures were
tested by LC/MS and molecular weights of glycosylamine and ARP
were confirmed (both 488 [M+H]+, m/z).
Results for disaccharides: The condensation of bisoprolol fuma-
rate and lactose (anhydrous and monohydrate) occurs with similar
rate for both excipients and was estimated by measuring decrease
in the concentration of the active substance. Large amounts of a
glycosylamine and ARP were observed in the samples with lactose
and maltose monohydrates. Reaction was much slower when
 M. Szalka et al. / European Journal of Pharmaceutical Sciences 59 (2014) 1–11 7

Fig. 4. The mechanism of the Maillard reaction and Amadori rearrangement of bisoprolol and glucose in acidic conditions.

anhydrous lactose was used; this confirms that water accelerates

the Maillard reaction. The rate of formation of N-formylbisoprolol
was the fastest for lactose monohydrate from all the excipients
used in the screening studies. Comparing obtained results we
observed that maltose monohydrate behaves very similar to lac-
tose monohydrate. All solid mixtures were also analysed by LC/
MS. Peaks due to glycosylamine with m/z 650 [M+H]+ and ARP with
m/z 650 [M+H]+ were found. Structure of N-formylbisoprolol was
also confirmed by LC/MS studies 354 [M+H]+ was found, retention
time of N-formylated bisoprolol was corresponding to the reten-
tion time of a synthesised standard.
Water content was determined by the Karl Fischer titration in
all samples used for the screening studies at the beginning of the
experiment. Obtained results were as follows: lactose anhydrous
0.12%, lactose monohydrate 5.24%, maltose monohydrate 5.35%,
glucose monohydrate 8.82%. The rate of the Amadori rearrange-
ment increased with increment of the water content in the tested
samples. Strong browning of the solid mixtures was also observed
during heating. Increase in UV–Vis absorbance of test solutions did
Fig. 5. Browning of the solid mixtures during heating monitored by the UV–Vis
not correlate with the increase in the concentration of glycosyl- spectrophotometry (at 490 nm).
amine, ARP, nor N-formylbisoprolol. No other known (according
to Ph. Eur.) or unknown impurities of bisoprolol fumarate were
observed during the HPLC analysis. Thus, it was assumed that browning phenomenon was assigned to the formation of
browning occurs due to the formation of melanoidins. Changes in melanoidines.
the absorbance of the test solutions are presented in Fig. 5. Blank
samples containing only appropriate carbohydrates were also 3.3. The stability studies of laboratory batches
placed in the screening studies to observe the carmelisation of sug-
ars. However, lactose and maltose remained white even after 6 h of Our next approach was to test the stability of solid mixtures and
heating in 95 °C. Glucose monohydrate turned into glassy trans- tablets, to simulate simple formulations of oral solid dosage forms
parent solid, browning of this mixture also was not observed. Since of medicinal product. Samples were packed into the polyethylene
carmelisation of sugars did not proceed in these conditions, the containers and stored for 6 months in pharmaceutically relevant
8 M. Szalka et al. / European Journal of Pharmaceutical Sciences 59 (2014) 1–11

conditions. Tested samples were placed in the climatic chambers
with monitored temperature and humidity, and analysed by HPLC
after 0, 3, and 6 months of storage. The accelerated conditions
(40 °C/75%RH) and the intermediate conditions (30 °C/65%RH)
according to ICH guidelines on the stability studies were selected.
The long-term studies (25 °C/60%RH) were not performed due to
the low reaction rate. Nevertheless, it is obvious that the Maillard
reaction occurs also in these conditions and large amounts of
N-formylbisoprolol may be found in the drug products, especially
at the end of a shelf-life period. Instead of the long-term conditions
storage in a laboratory oven in 50 °C/10%RH was applied. The list of
tablet fillers and lubricants used in this study was presented in the
experimental section. Sorbitol and calcium phosphate were used
for the comparison purposes, both excipients have no carbonyl
group, thus (in theory) are not substrates for the Maillard reaction.
At first the influence of the ratio of bisoprolol fumarate to the
excipient (glucose) was tested. For this purpose tablets containing
5%, 20%, and 40% of bisoprolol were prepared and placed in the cli-
matic chambers. The biggest amounts of glycosylamine, ARP, and
N-formylbisoprolol were found in samples containing 5% of biso-
prolol fumarate, then 20% and 40% in decreasing order. Similar
results were obtained in all conditions and for all carbohydrates.
It may be concluded that the ratio of a secondary amine to carbo-
hydrate is an important factor which determines the rate of the
Maillard reaction. Unfortunately, the strengths of drug products
containing bisoprolol used for the therapeutic purposes are rela-
tively small: tablets with 2.5, 5, and 10 mg of the active substance
are available on the market. Low content of bisoprolol (approx. 5%
of tablet weight) in relation to tablet filler promotes the Maillard
reaction in drug products. Our results stay in the conformance with
results obtained byMonajjemzadeh et al. (2009) in studies of the
Maillard reaction between baclofen and lactose.
The second approach included testing of the most popular fillers
used in a pharmaceutical industry on the rate of the Maillard reac-
tion. For this purpose mixtures containing 5% (w/w) of bisoprolol
fumarate and various excipients were prepared, magnesium stea-
rate was used in all samples as a lubricant to enable tableting.
Uncompressed solid mixtures and tablets were placed in the cli-
matic chambers and analysed after 0, 3, and 6 months. In all tested
samples the reaction rate was faster in the compressed tablets than
in the loose powders, this phenomenon was observed indepen-
dently on the carbohydrate type. Compression improves the
physical contact between particles which effects in a faster con-
densation of the carbonyl and amine functional groups. Tablet
hardness is another important factor which influences the Maillard
reaction. It is important to highlight that the Maillard reaction pro-
ceeds also in uncompressed solid mixtures (in the accelerated con-
ditions), thus, it may also occur in capsules containing bisoprolol
fumarate and reducing carbohydrate.
Filler type was the most critical parameter which affected the
rate of the Maillard reaction in our studies. The fastest degradation
occurred in the samples with glucose and was simply explained by
high molar ratio of excipient to bisoprolol. Only 64% and 89% of the
remaining bisoprolol was determined in the samples stored for
6 months in 50 °C and 40 °C, respectively. 6.5% of N-formylbisopro-
lol was found in the former sample, this quantity exhibits the qual-
ification level of the identified impurity over 40 times (according to
the EMEA guidelines). Large amounts of the ARP were observed
only when glucose was used as filler, for the other excipients this
intermediate reaction product was found only in small quantities
or was not found at all, even when N-formylated bisoprolol was
detected. Preparation of the laboratory batches did not include
film-coating of tablets to avoid the introduction of water or organic
solvents. It was assumed that uncoated tablets are not well pro-
tected from air, thus oxidative cleavage occurs readily after the for-
mation of ARP and leads to N-formylbisoprolol. Comparison of the
results obtained for tablets with disaccharides revealed that not
only temperature but also humidity in the climatic chambers has
strong effect on the Maillard reaction. In a laboratory oven (RH
approx. 10%) the samples with maltose and lactose monohydrates
had much faster reaction rate than a sample with lactose anhy-
drous. On the other hand the reaction rates for all three excipients
are very similar when stored in a high humidity chamber (RH 75%),
leading to the conclusion that there is no advantage of using anhy-
drous lactose in these conditions. Polyethylene bottles were used
as the immediate package in our stability studies, whereas usually
drug products are packed in the Al/PVC blisters which guarantee
better protection from moisture. It should be considered that using
a better package material could unveil the advantage of anhydrous
lactose over monohydrate.
When glucose, maltose, or lactose monohydrates were used as
excipients an intensive browning and increase of moisture were
observed visually. Also significant loss in the assay of bisoprolol
fumarate was determined by the HPLC analysis, which was not
compensated by the increase in impurities, this is why results of
assay and impurities do not sum to 100% in Table 1. It is doubtful
that response factor of the active substance and its related sub-
stances differ that much, thus, explanation of the formation of mel-
anoidins was applied basing on the fact that intensive browning
occurred. Melanoidins are nitrogenous co-polymers and polymers

Table 1
The stability study results (±standard deviation) for tablets containing 5% of bisoprolol fumarate and various excipients stored in the accelerated conditions (40 °C/75%RH) for
6 months.

Storage time Glucose Maltose Lactose Lactose Sorbitol (%) Calcium hydrogen
monohydrate (%) monohydrate (%) monohydrate (%) anhydrous (%) phosphate (%)
Initial
Remaining bisoprolol 99.7 ± 0.3 101.0 ± 0.1 98.0 ± 0.2 97.1 ± 0.1 98.8 ± 0.5 100.8 ± 0.2
Glycosylamine 0.00 0.00 0.00 0.00 0.00 0.00
ARP 0.00 0.00 0.00 0.00 0.00 0.00
N-formylbisoprolol 0.00 0.00 0.00 0.00 0.00 0.00
3 months
Remaining bisoprolol 92.1 ± 0.3 91.9 ± 3.3 96.2 ± 0.1 96.6 ± 0.3 97.1 ± 0.1 103.2 ± 0.2
Glycosylamine 0.12 ± 0.01 0.05 ± 0.01 0.03 ± 0.02 0.00 0.00 0.00
ARP 0.89 ± 0.02 0.30 ± 0.07 0.00 0.03 ± 0.01 0.00 0.00
N-formylbisoprolol 0.30 ± 0.01 0.19 ± 0.00 0.13 ± 0.00 0.15 ± 0.00 0.02 ± 0.00 0.00
6 months
Remaining bisoprolol 89.6 ± 1.5 79.2 ± 1.7 97.5 ± 0.1 102.2 ± 0.3 96.5 ± 0.6 100.1 ± 0.1
Glycosylamine 0.42 ± 0.03 0.17 ± 0.01 0.12 ± 0.00 0.16 ± 0.01 0.03 ± 0.00 0.00
ARP 2.72 ± 0.02 0.22 ± 0.01 0.00 0.00 0.00 0.00
N-formylbisoprolol 0.82 ± 0.00 0.29 ± 0.01 0.29 ± 0.01 0.27 ± 0.00 0.03 ± 0.00 0.00
 M. Szalka et al. / European Journal of Pharmaceutical Sciences 59 (2014) 1–11 9

which are considered as the group of final products of the Maillard

reaction. Melanoidins are large molecules and cannot be deter-
mined by the HPLC analysis. Martin et al. (Hammouda and
Salakawy, 1971) proved that browning is a function of the concen-
tration of melanoidins. The dark-brown slurry residue was found
instead of tablets in a polyethylene bottle containing glucose and
bisoprolol after storage for 6 month in 50 °C; the browning and for-
mation of the melanoidins may be a rapid process.
The samples with cellulose microcrystalline or maize starch
were relatively stable, approximately 100% of bisoprolol fumarate Fig. 6. Proposition for the complex of magnesium ion and glycosylamine formed in
was determined in those samples at the end of the stability studies the samples with magnesium stearate.

in all conditions. As it was suspected both polysaccharides are inert

to the Maillard reaction because of a small number of terminal car- the catalysing effect of magnesium stearate and assigned it to
bonyl groups. localised changes in the pH of solid mixtures. Monajjemzadeh
Sorbitol is a sugar alcohol which may be obtained by reducing et al. (2009) and Qiu et al. (2005) also refer to Wirth’s studies,
the carbonyl group of glucose into hydroxyl group. Sorbitol should although his considerations have rather hypothetical character.
be inert to the Maillard reaction, however, analysis of tablets In our opinion there may be other explanation of this phenomenon.
stored in 50 °C revealed presence of small amounts of glycosyl- Magnesium ions (2+) form stable five member rings with the glyco-
amine and N-formylbisoprolol. N-formylated bisoprolol was also sylamine imitating O-protonation, as shown in Fig. 6. These weak
found in the samples stored for 6 months in the accelerated condi- forces catalyse further rearrangement in the presence of large
tions. As it was previously explained commercially available sorbi- amounts of water which explains why reaction is favoured in a
tol may contain unreacted glucose, which undergoes the Maillard high humidity chamber whereas does not proceed well in a labora-
reaction in tablets. tory oven. In the accelerated conditions stearic acid had much
Finally, calcium hydrogen phosphate was the best filler for smaller catalysing effect on the Maillard reaction than in a low
tablets containing bisoprolol fumarate. This excipient is not a sub- humidity chamber. It was assumed, that water acts as a proton
strate for the Maillard reaction, thus, formation of the glycosyl- acceptor, this decrease O-protonation of glycosylamine and inhib-
amine, ARP, and N-formylated bisoprolol was not observed. No its the rearrangement.
significant degradation of the active substance occurred, 100% of Finally, in the accelerated conditions the reaction occurs much
assay determined by the HPLC analysis confirmed that this formu- slower when sodium stearyl fumarate, talc, and silica were used.
lation is very stable. The summary of results obtained for fre- These lubricants should be considered by formulation scientists
quently used excipients stored for 6 months in the accelerated to prevent the Maillard reaction. The influence of various lubri-
conditions is presented in Table 1. cants on the rate of the formation of N-formylbisoprolol is pre-
The type of a lubricant used in the manufacturing process is also sented in Fig. 7.
an important factor which affects the rate of the Maillard reaction.
The most frequently used lubricants were tested: magnesium stea- 3.4. The stability studies of commercially available products
rate, stearic acid, talc, sodium stearyl fumarate, and silica colloidal.
Solid mixtures and tablets containing 20% (w/w) of bisoprolol The 6-month stability studies in the accelerated conditions
fumarate and 79% lactose monohydrate with addition of 1% of a (40 °C/75%RH) were conducted to confirm that the Maillard reac-
tested lubricant were prepared and stored for 6 months in a labo- tion occurs also in commercially available products. For this pur-
ratory oven (50 °C/10%RH) and in the accelerated and intermediate pose two popular drug products containing bisoprolol fumarate
conditions. One blank sample without the addition of a lubricant (2.5 mg strength) and lactose monohydrate were chosen and
was prepared for the comparison purposes; in this sample lubri- placed in a climatic chamber. Both products contain magnesium
cant was substituted by the equivalent amount of lactose monohy- stearate as a lubricant. Film-coated tablets were kept in the origi-
drate. The laboratory batches were removed from the climatic nal marketing package (Al/PVC foil blister placed in a cardboard
chambers and analysed by HPLC after 0, 3, 6 months of storage.
N-formylbisoprolol was the main degradation product found in
all tested samples, whereas glycosylamine and ARP were found
only at very low levels or were not detected at all. The reaction
was monitored by measuring the concentration of the final reac-
tion product. Results obtained for the solid mixtures and tablets
were very similar, thus, in further text we will refer only to the
uncompressed solid powders to avoid the influence of a physical
compression. In the low humidity laboratory oven (50 °C/10%RH)
0.24% of N-formylated bisoprolol was observed when stearic acid
was used as a lubricant. The reaction rate was 15 times faster than
in a blank sample, this observation may be attributed to the acidic
micro-environment created by the lubricant. The addition of stea-
ric acid catalyses the Amadori rearrangement by O- or N-proton-
ation. Other tested lubricants had no significant influence on the
degradation of the active substance in low moisture environment.
However, the analysis of mixtures stored in high humidity cham-
ber (40 °C/75% RH) revealed that in these conditions magnesium
stearate has the strongest catalysing effect. About 0.06% of
N-formylated bisoprolol was found in this sample, whereas only
0.01% and 0.04% were found in a blank sample and in mixture with Fig. 7. Increase of N-formylbisoprolol in the samples with various lubricants during
stearic acid, respectively. Wirth et al. (Ledl et al., 1986) observed 6 month storage (50 °C/10%RH).
10 M. Szalka et al. / European Journal of Pharmaceutical Sciences 59 (2014) 1–11
Table 2
The amount of N-formylbisoprolol and assay of active substance (± standard deviation) determined in commercially available products during the accelerated stability studies.
Time of storage (month) Content of N-formylbisoprolol (%)
Brand A
0 0.04 ± 0.02
3 0.14 ± 0.01
6 0.20 ± 0.02
Fig. 8. Increase of N-formylbisoprolol concentration in commercially available
products after storage for 6 months in the accelerated conditions.
box) to simulate real storage conditions. In further text drug
products were marked as brand A and brand B. Similarly to our
laboratory batches (with lactose) no glycosylamine or 1-deoxy-1-
amino-2-ketose were detected in both commercial products.
However, significant amounts of N-formylbisoprolol were
observed in all samples after 3 and 6 months of the stability stud-
ies. Brand A contained N-formylated bisoprolol even at 0 time
point, which means that the Maillard reaction may occur in tablets
supplied by a pharmacy. LC/MS studies confirmed the molecular
weight 354 [M+H]+ (m/z) for N-formylbisoprolol in both generic
drug products. Table 2 and Fig. 8 present comparison of the stabil-
ity results obtained for brand A and B. The full comparison of com-
mercial products is difficult due to the lack of knowledge about the
manufacturing process, for example duration of a film-coating pro-
cess may increase the water content which improves the rate of
the Maillard reaction.
4. Conclusions
The formulation incompatibility between bisoprolol fumarate
and various reducing carbohydrates used as tablet fillers was
presented in this paper. HPLC–UV/MS analysis confirmed the
presence of a glycosylamine, 1-deoxy-1-amino-2-ketose, and
N-formylbisoprolol. It indicates that the Maillard reaction followed
by the Amadori rearrangement occurs readily in pharmaceutical
preparations. The solid screening studies were carried out to test
the influence of water on the reaction rate. Laboratory batches of
bisoprolol fumarate with various diluents were prepared, the fast-
est degradation was observed for glucose indicating that high
molar ratio of the excipient to the active substance plays important
role in the degradation. The reaction rate was much faster for
monohydrates than for anhydrous fillers. The excipient of choice
for bisoprolol fumarate is calcium hydrogen phosphate which is
not a substrate for the Maillard reaction, no increase in individual
Assay of bisoprolol fumarate (%)
Brand B Brand A Brand B
0.00 99.6 ± 0.1 99.9 ± 0.1
0.17 ± 0.02 98.6 ± 0.2 99.2 ± 0.1
0.38 ± 0.03 98.0 ± 0.1 98.5 ± 0.1
or total impurities was observed, drug product maintained its ini-
tial potency. Five common lubricants used in the pharmaceutical
industry were tested, from which magnesium stearate and stearic
acid proved to increase the reaction rate in tablets and solid mix-
tures. Both lubricants should be avoided in the pharmaceutical
preparations which contain amines and reducing carbohydrates.
The complex of magnesium ions and glycosylamine accelerates
the Amadori rearrangement.
In all the samples with reducing carbohydrates N-formylbisoprolol
was identified as the final reaction product. The N-formylated amines
are usually stable, may be easily synthesised and used as reference
standards. The intensive browning of solid mixtures and tablets was
attributed to the formation of melanoidins. There are information in
the available literature that Maillard reaction products may be
genotoxic, cancerogenic, or cytotoxic (Delgado-Andrade et al., 2010;
Zabala-Diaz et al., 2010). However, detailed toxicological studies for
bisoprolol-related impurities should be carried out to confirm
biological activity. N-formylbisoprolol was found in two commer-
cially available products indicating that the Maillard reaction occurs
even in tablets stored in the original marketing package. According
to the EMEA and FDA guidelines impurities at high levels should be
identified and qualified to assure the safety of a medicinal
preparation. The safety of the drug products may be also altered by
significant loss of the potency. It is advised to avoid the combination
of bisoprolol fumarate with reducing carbohydrates in tablet and
capsule formulations.
Acknowledgment
We are grateful to PhD Edward Rokaszewski for valuable infor-
mation and technical advice on the studies of the Maillard reaction
in pharmaceutical preparations.
References
Amadori, M., 1925. Atti. Accad. Lincei, 6, 2, 337 (1925); 9, 68, 226 (1929); 13, 72, 95
(1931).
Castello, R.A., Mattock, A.M., 1962. Discoloration of tablets containing amines and
lactose. J. Pharm. Sci. 51, 106–108.
Delgado-Andrade, C., Morales, F., Seiquer, I., Navarro, M., 2010. Maillard reaction
products profile and intake from Spanish typical dishes. Food Res. Int. 43, 1304–
1311.
European Medicines Agency, 2003. Guideline on Stability Testing. Stability Testing
of Existing Active Substances and Related Finished Products. CPMP/QWP/122/
02, rev 1 corr, 2003.
Gottschalk, A., 1952. Biochem. J. 52, 455.
Hammouda, V., Salakawy, S.A., 1971. Lactose induced discoloration of amino drugs
in solid dosage form. Pharmazie 26, 181.
Hodge, J.E., 1955. The Amadori rearrangement. Adv. Carbohydr. Chem. 10, 169–205.
Hodge, J.E., Rist, C.E., 1953. The Amadori rearrangement under new conditions and
its significance for non-enzymatic browning reactions. J. Am. Chem. Soc. 75,
316–322.
Isbell, H.S., Frush, H.L., 1958. J. Res. Natl. Bur. Std. 39; C.A. 52, 15432.
Kawakishi, S., Cheng, R., 1994. Novel decomposition of Amadori compound
catalyzed by copper ion. J. Agric. Food Chem. 42, 700–703.
Kuhn, R., Dansi, A., 1936. A molecular rearrangement of N-glucosides. Ber. Dtsch.
Chem. Ges. 69B, 1745–1754.
Kuhn, R., Weygand, F., 1937. The Amadori rearrangement. Ber. Dtsch. Chem. Ges.
70B, 769–772.
Labuza, T.P., Baisier, W.M., 1992. The kinetics of nonezymatic browning. In:
Schwartzberg, H.G., Hartel, R.W. (Eds.), Physical Chemistry of Foods, seventh
ed. Marcel Dekker, New York.
 M. Szalka et al. / European Journal of Pharmaceutical Sciences 59 (2014) 1–11
Labuza, T.P., Massaro, S.A., 1990. Browning and amino-acid loss in model total
parental-nutrition solutions. J. Food Sci. 55, 821–826.
Ledl, F., 1990. Chemical pathways of the Maillard reaction. In: Finot, P.A.,
Aeschbacher, H.U., Hurrell, R.F., Liardon, R. (Eds.), The Maillard Reaction in
Food Processing Human Nutrition and Physiology. Birkhauser Verlag, Basel, pp.
19–42.
Ledl, F., Schleicher, E., 1990. New aspects of the Maillard reaction in foods and in the
human body. Angew. Chem., Int. Ed. 29 (6), 565–594.
Ledl, F., Fritsch, G., Hiebl, J., Pachmayer, O., Severin, T., 1986. Degradation of Maillard
products. In: Fujimaki, M., Namiki, M., Kato, H. (Eds.), Amino-Carbonyl
Reactions in Food and Biological Systems. Elsevier, Amsterdam, pp. 173–182.
Maillard, L.C., 1912. Action des acides amines sur les sucres: formation des
melanoidines par voie methodique. C. R. Acad. Sci. Ser. 2 (154), 66–68.
Martins, S., Van Boekel, M., 2003. Melanoidins extinction coefficient in the glucose/
glycine Maillard reaction. Food Chem. 83, 135–142.
McMurray, J.J., Adamopoulos, S., Auricchio, A., Bohm, M., Dickstein, K., et al., 2012.
ECS Guidelines for the diagnosis and treatments of acute chronic and heart
failure: the task force for the diagnosis and treatment of acute and chronic heart
failure 2012 of the European Society of Cardiology. Eur. Heart J. 33 (14), 1787–
1847.
Merck KGaA Darmstadt Germany, 2001. Bisoprolol, Cardioselective Beta-Blocker.
<www.bisoprolol.com>.
Mills, F.D., Baker, B.G., Hodge, J.E., 1970. Thermal degradation of 1-deoxy-1-
piperidino-d-fructose. Carbohydr. Res. 15, 205–213.
Mitts, E., Hixon, R.M., 1944. J. Am. Chem. Soc. 66, 483.
Monajjemzadeh, F., Hassanzadeh, D., Valizadeh, H., Siahi-Shadbad, M.R., Mojarrad,
J.S., Robertson, T., Roberts, M.S., 2009. Assessment of feasibility of Maillard
reaction between baclofen and lactose by liquid chromatography and tandem
mass spectrometry, application to pre formulation studies. AAPS PharmSciTech
10, 649–659.
View publication stats
11
Qiu, Z., Stowell, J.G., Morris, K.R., Byrn, S.R., Pinal, R., 2005. Kinetic study of the
Maillard reaction between metoclopramide hydrochloride and lactose. Int. J.
Pharm. 303, 20–30.
Rowe, R.C., Sheskey, P.J., Weller, P.J., 2003. Handbook of Pharmaceutical Excipients,
fourth ed. Pharmaceutical Press, American Pharmaceutical Association.
Serajuddin, A.T., Thakur, A.B., Ghoshal, R.N., Fakes, M.G., Ranadive, S.A., Morris, K.R.,
et al., 1999. Selection of solid dosage form composition through drug-excipient
compatibility testing. J. Pharm. Sci. 88, 696–704.
Simon, H., Kraus, A., 1970. Mechanistische Untersuchungen über Glycosylamine,
Zuckerhydrazone, Amador-Umlagerungsproducte und Osazone. Fortschr.
Chem. Forsch. 14, 430–470.
Thumma, S., Repka, M.A., 2009. Compatibility studies of promethazine
hydrochloride with tablet excipients by means of thermal and non-thermal
methods. Pharmazie 64, 183–189.
Tjan, S.B., Ouweland, G.A.M.V.D., 1974. PMR investigation into the structure of some
n-substituted 1-amino-1-deoxy-D-fructoses (Amadori rearrangement
products): evidence for a preferential conformation in solution. Tetrahedron
30, 2891–2897.
Wang, J., Lu, Y.-M., Liu, B.-Z., He, H.-Y., 2008. Electronspray positive ionization
tandem mass spectrometry of Amadori compounds. J. Mass Spectrom. 43, 262–
264.
Weygand, F., Bergmann, A., 1947. Chem. Ber. 80, 261.
Wirth, D.D., Baertschi, S.W., Johnson, R.A., Maple, S.R., Miller, M.S., Hallenbeck, D.K.,
Gregg, S.M., 1998. Maillard reaction of lactose and fluoxetine hydrochloride, a
secondary amine. J. Pharm. Sci. 87, 31–39.
Wolfrom, M.L., Schuetz, R.D., Cavalieri, L.F., 1949. J. Am. Chem. Soc. 71, 3518.
Zabala-Diaz, I., Chalova, V., O’Bryan, C., Crandall, P., Ricke, S., 2010. Effect of soluble
Maillard reaction products on cadA expression in Salmonella typhimurium. J.
Environ. Sci. Health B45, 162–166.