Accelerated Forced Degradation of Pharmaceuticals in Levitated Microdroplet Reactors

Article abstract of DOI:10.1002/chem.201801176

Forced degradation is a method of studying the stability of pharmaceuticals in order to design stable formulations and predict drug product shelf life. Traditional methods of reaction and analysis usually take multiple days, and include LC-UV and LC-MS product analysis. In this study, the reaction/analysis sequence was accelerated to be completed within minutes using Leidenfrost droplets as reactors (acceleration factor: 23–188) and nanoelectrospray ionization MS analysis. The Leidenfrost droplets underwent the same reactions as seen in traditional bulk solution experiments for three chemical degradations studied. This combined method of accelerated reaction and analysis has the potential to be extended to forced degradation of other pharmaceuticals and to drug formulations. Control of reaction rate and yield is achieved by manipulating droplet size, levitation time and whether or not make-up solvent is added. Evidence is provided that interfacial effects contribute to rate acceleration.

Full text of DOI:10.1002/chem.201801176

A Journal of
Accepted Article
Title: Accelerated Forced Degradation of Pharmaceuticals in Levitated
Microdroplet Reactors
Authors: Yangjie Li, Yong Liu, Hong Gao, Roy Helmy, W Peter
Wuelfing, Christopher Welch, and R Graham Cooks
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Chem. Eur. J. 10.1002/chem.201801176
Link to VoR: http://dx.doi.org/10.1002/chem.201801176
Supported by

10.1002/chem.201801176
Chemistry - A European Journal
FULL PAPER
Accelerated Forced Degradation of Pharmaceuticals in Levitated
Microdroplet Reactors
Yangjie Li, Yong Liu, Hong Gao, Roy Helmy, W. Peter Wuelfing, Christopher J. Welch and R. Graham
Cooks*
Abstract: Forced degradation is a method of studying the stability of
pharmaceuticals in order to design stable formulations and predict
drug product shelf life. Traditional methods of reaction and analysis
usually take multiple days, and include LC–UV and LC–MS product
analysis. In this study, the reaction/analysis sequence was
accelerated to be completed within minutes using Leidenfrost
is critical to understand potential liabilities in short time frames
rather than waiting for several weeks for an actual formulation to
degrade sufficiently for measurement.[2b] The study presented
here explores new methods to rapidly assess degradation
chemistry of APIs under dehydration, oxidation and hydrolysis
conditions in solution.
droplets as reactors (acceleration factor: 23
–
188) and
It is known that the rates of chemical reactions are often
accelerated in small droplets or thin films vs. the corresponding
bulk or larger droplet rates.[3] Concentration and pH changes
during solvent evaporation and incomplete solvation of
molecules at the air-droplet interface are the major factors that
contribute to such reaction acceleration.[3a] One method of
droplet acceleration employs the Leidenfrost effect to create
small levitated droplets with no net charge. By pouring a liquid
onto a surface at a temperature significantly greater than the
boiling point of the liquid, levitating droplets can be created from
which solvent gradually evaporates. A previous study of several
reactions showed acceleration in Leidenfrost droplets vs. bulk by
factors of 2 – 50, as judged by the time required to reach a
particular fractional conversion of the reactant to product.[4] The
millimeter-sized Leidenfrost droplets last for some minutes and
when used as reaction vessels for APIs they force degradation
to occur much faster than would be the case under traditional
‘forced degradation’ conditions as is now shown.
Compared to the traditional forced degradation, not only is
the reaction accelerated by a factor of one to two orders of
magnitude but the subsequent analysis by non-accelerating[3a]
nanoelectrospray ionization mass spectrometry (nESI–MS) is
faster than traditionally used LC–UV and LC–MS methods (Fig.
1). The Leidenfrost experiment involves reactions in confined
volumes under conditions similar to those encountered in
ambient ionization.[5] Well-studied degradation chemistry was
chosen to test this methodology: acid degradation of tetracycline
(TCN) and hydrochlorothiazide (HCTZ) and oxidative
degradation of trifluoperazine (TFP).[6] We followed the
degradation in Leidenfrost droplets, while maintaining nearly
constant volumes, and used nESI–MS for analysis. The data
were compared with those for degradation in the corresponding
bulk solution (traditional forced degradation) under similar
conditions with respect to the concentration of APIs and
degradation reagents, composition of the solvent, and reaction
temperature.
nanoelectrospray ionization MS analysis. The Leidenfrost droplets
underwent the same reactions as seen in traditional bulk solution
experiments for three chemical degradations studied. This combined
method of accelerated reaction and analysis has the potential to be
extended to forced degradation of other pharmaceuticals and to drug
formulations. Control of reaction rate and yield is achieved by
manipulating droplet size, levitation time and whether or not make-
up solvent is added. Evidence is provided that interfacial effects
contribute to rate acceleration.
Forced degradation studies have long been part of the
pharmaceutical development process for active pharmaceutical
ingredients (APIs). The aim of such studies is to selectively
probe chemical reactivity of active compounds so as to
understand the chemical degradation that is possible in drug
formulations. Further, the rates of chemical reactions in these
studies yield some insight into drug product shelf life. Once
possible degradation mechanisms are understood, formulation
strategies can be developed to minimize chemical degradation
and enable design of a stable formulation product.[1]
Even the established forced degradation methods are
time-consuming, typically being performed in solution over the
course of 1 – 7 days for each API under each chosen condition,
even when the extent of degradation is limited to 10% – 20%.[1d]
A recent paper by Dow et al. describes forced stress workflows,
including how and when structures are assessed in accordance
with regulatory guidance.[2a] Acceleration of degradant formation
[*]
Y. Li, Prof. R. G. Cooks
Department of Chemistry, Purdue University
560 Oval Drive, West Lafayette, IN 47907 (USA)
E-mail: cooks@purdue.edu
Homepage: http://aston.chem.purdue.edu
Dr. Y. Liu, Dr. R. Helmy, Dr. W. P. Wuelfing,
Dept. of Analytical Sciences
MRL, Merck & Co., Inc.
West Point, PA 19446 (USA)
Dr. H. Gao, Dr. C. J. Welch,
Dept. of Analytical Research & Development
MRL, Merck & Co., Inc.
We performed traditional forced degradation by preparing
a stock stress solution of hydrochloric acid or hydrogen peroxide,
and a stock solution of the API of interest. Then the solutions
were mixed and allowed to react at a set temperature (see
Scheme 1 and Supporting Information, Sec.1 for details) for a
set period of time. Likewise, the Leidenfrost reaction mixture
was prepared the same way but instead of allowing the mixture
to age at a fixed temperature, it was dropped into a concave
ceramic well which was placed on a hotplate to form a single
Rahway, NJ 07065 (USA)
Supporting information for this article is given via a link at the end of
the document.
This article is protected by copyright. All rights reserved.

10.1002/chem.201801176
Chemistry - A European Journal
FULL PAPER
Figure 1. Comparison of (a) traditional and (b) accelerated reaction/analysis sequence. When make-up solvent is added continuously, the Leidenfrost
acceleration is purely a result of the increased rate constant in the droplet reactions.
Considering the suggested mechanism of reaction
acceleration in droplets,[3a] the microdroplet reactors were held
at nearly constant volume, so that concentration and pH
changes due to evaporation could be eliminated to allow direct
comparison of the rate constants for Leidenfrost to traditional
bulk experiments. This was realized by automatically adding
make-up solvent to compensate for that lost during evaporation.
The solvent was added via a syringe pump at a flow rate which
depended on the size and evaporation rate of the levitating
droplet (see Supporting Information, Table S1 for details).
In this study, the compositions of the final reaction
mixtures in both bulk and Leidenfrost were analyzed by nESI–
MS for direct comparison and the conversion ratio (CR), the ratio
of the ion intensities of degradants to the sum of the intensities
of degradants and the API, was used to infer the extent of
reaction. The simple measurement of the time required to
achieve the same conversion ratio is a good measure of the
acceleration factor,[4] while kinetic measurements[7] were
performed in this study for a more precise measure of the
acceleration factor, as summarized in Table 1.
Table 1. Degradation of APIs under Leidenfrost vs. conventional forced
degradation conditions and the reaction acceleration factor
Scheme 1. Three degradation reactions studied, showing the compositions of
Reaction
Acceleration
factor [c]
API and degradation
reagent [a]
Slope
(Leidenfrost) [b]
the solutions used in both the bulk and Leidenfrost experiments. The major
ionic forms of the APIs and their degradants and the nominal mass/charge
ratios of products and reagents measured by MS are indicated. The structure
Slope (bulk)
of 2a was elucidated by tandem mass spectrometry, Fig. S1, as
representative example.
a
TCN (1), HCl
TFP (2), H2O2
HCTZ (3), HCl
0.000187 min-1
0.000323 min-1
0.000318 min-1
0.0352 min-1
0.00762 min-1
0.00725 min-1
188
23.6
23.6
levitated Leidenfrost droplet. Reaction occurred in the levitated
droplet. Note that although a high temperature was used to heat
the plate, the temperature of the levitating solution was very
much lower and comparable to that used in the corresponding
bulk experiment using traditional forced degradation. (The
Leidenfrost droplet was levitated but not boiled and its
temperature is roughly estimated as 20 °C less than the boiling
point of the solvent.[4])
[a]
Identical solutions were used for bulk and Leidenfrost experiments.
[b] The slopes of the trend lines in both bulk and Leidenfrost kinetic profiles
(CR vs time) have units of min-1 and the ratio of the slopes (Leidenfrost
over bulk rate) was used to represent the reaction acceleration factor.
[c] Leidenfrost reactions in acidic conditions were done in 500 µL droplets
while oxidative degradation was carried out in 100 µL droplets. The
acceleration factor value should not be used to compare these
experiments since reaction rate in droplet reactor depends on the size
(See Supporting Information, Sec. 4).
This article is protected by copyright. All rights reserved.

10.1002/chem.201801176
Chemistry - A European Journal
FULL PAPER
Oxidative degradation of trifluoperazine (2) is discussed as
a proof-of-concept experiment here. Comparison of the kinetic
profiles in bulk solution and in Leidenfrost droplets confirms that
the degradation occurs at very different rates (Fig. 2). Each data
point represents information from at least three trials. The y-
intercepts are slightly greater than zero since in both
experiments timing started several minutes after the reactants
were mixed.
The
traditional
and
Leidenfrost
accelerated
reaction/analysis sequences of two other degradations were
also compared. The kinetic profiles of acid degradation of
tetracycline and hydrochlorothiazide, plotted in Fig. S2 and S3,
were used to calculate reaction acceleration factors of 188 and
23.6, respectively. The MS recorded in both experiments
showed no significant differences with respect to APIs and their
major degradants in the Leidenfrost and bulk experiments
except for the time required to achieve particular degradation
levels.
To gain deeper insight into the mechanism of reaction
acceleration, the oxidative degradation of trifluoperazine was
carried out in a microdroplet reactor using different droplet sizes
but maintaining constant sizes during the course of reaction.
This procedure eliminates concentration effects as variables and
allows the effects of interfacial factors to be studied directly. The
data (Fig. 3) show that the reaction acceleration factor correlates
inversely with the size of the Leidenfrost droplet and hence is
positively correlated to the surface/volume ratio of the levitated
droplets. The surface areas and volumes of the droplets are
estimates only (see SI, Sec. 4) but the data show that the
reaction acceleration factor for the smaller droplets (ca. 100 µL)
is greater than that for the 500 µL droplets. The size-dependent
reaction acceleration factor in constant volume levitated
[3a]
microdroplets is consistent with the proposal
that the
solution/air interface is key to reaction acceleration. The result
also suggests that the ease of manipulation of reaction rate in
the Leidenfrost acceleration experiment makes it an even more
attractive acceleration method for forced degradation, where
control of degradation within the range of 10% – 20% is usually
needed.
Figure 2. Conversion ratio (CR) of TFP (2) to 2a and 2b over time in (a) bulk
and (b) constant volume 100 µL Leidenfrost droplet reactor. Mass spectra are
for the same conversion ratio, calculated as the sum of the peak intensities of
protonated 2a and 2b (the major oxidative degradants seen in MS) over the
sum of the peak intensities for the ions of 2, 2a and 2b.
Besides the similarity in the kinetic profiles, the mass
spectra of the reaction mixtures were recorded at particular
times chosen to correspond to similar extents of degradation.
These data show no significant differences, which confirms the
reliability of using Leidenfrost droplet reactors to accelerate this
particular degradation. Note that this method uses relative
abundance (RA) ratios, i.e. peak height ratios in mass spectra,
and does not account for matrix effects (which are similar in the
two solutions) or for ionization efficiencies (intrinsic properties of
the two species being measured). It also does not account for
differences (if any) in by-products associated with the different
experimental conditions. This quick analytical method provides
insight into the extent of degradation without the need for any
separation.
Figure 3. Conversion ratio (CR) of TFP (2) to 2a and 2b over time: (circles)
constant volume 100 µL Leidenfrost droplet reactor; (triangles) constant
volume 500 µL Leidenfrost droplet reactor.
Tandem mass spectrometry (MS/MS) was used (Fig. S1)
to elucidate the structure of mono-oxidized degradant 2a as the
S-oxide produced under the reaction conditions described in this
study. A fragment ion m/z 141 indicates that the piperazine ring
in 2a was not oxidized, while the appearance of ions of m/z 296
and m/z 324, both shifted by 16 compared to the corresponding
phenothiazine ring containing fragments in protonated 2,
confirms S-oxidation. The data show that nESI–MS can be
adequate as a rapid and reliable method for both analyzing the
extent of degradation and for characterizing the degradant.
The kinetic profile of the constant volume Leidenfrost
experiments shows that reaction yield can be increased by
levitating a Leidenfrost droplet for a longer time. In comparison,
Leidenfrost experiments examined without adding make-up
solvent (Fig. 4) showed very rapid product formation. In one
such case, the 500 µL oblate spheroidal droplet was collected
when it had evaporated to ca. 30 µL (just before complete
evaporation), diluted and analyzed. Oxidative degradation
assessed by MS was found to be extensive during the
evaporation and shrinking of the droplet. After reacting for ca. 60
This article is protected by copyright. All rights reserved.

10.1002/chem.201801176
Chemistry - A European Journal
FULL PAPER
s, the major axis of the oblate spheroidal droplet had shrunk
from ca. 13 mm to 5 mm (about 100 µL, see Supporting
Information Sec. 4 for estimation method), and the conversion
ratio was measured to be 5%. In a shrinkage experiment of 90 s,
the major axis of the droplet had shrunk to ca. 3 mm (about 30
µL), and the mass spectrum showed that 68% of TFP has been
degraded (Fig. 4b). These results suggest that letting the droplet
shrink without adding make-up solvent and collecting the droplet
just before all solvent has evaporated is potentially a quick, very
rough synthetic method (compare ref. 3, 4). Parenthetically we
note that the use of shrinking levitated droplets for synthesis was
seen for acid dehydration of TCN. About 70 s after generating
the 500 uL droplet and letting it evaporate without adding make-
up solvent, the final ca. 10 uL droplet was collected, diluted, and
analyzed. The spectrum (Fig. 4d) showed a 99% conversion
ratio to give a pure product.
Acknowledgements
This work was supported by Merck Sharp & Dohme Corp. a
subsidiary of Merck & Co., Inc., Kenilworth, NJ, USA, under a
Master Agreement with Purdue University on Chemical
Instrumentation. Project number LKR161299. We thank Marcela
Nefliu and Andreas Abend for valuable technical advice.
Conflict of interest
The authors declare no conflict of interest.
Keywords: forced degradation • kinetics • Leidenfrost effect •
mass spectrometry • reaction acceleration
[1]
a) International conference on harmonisation, guidance for industry,
stability testing of new drug substances and products Q1A (R2); b) B. A.
Olsen, et al., Impurity investigations by phases of drug and product
development,
Trends
in
Analytical
Chemistry
(2017),
https://doi.org/10.1016/j.trac.2017.10.025; c) S. W. Baertschi, P. J.
Jansen, K. M. Alsante in Pharmaceutical Stress Testing: Predicting
Drug Degradation, 2nd Ed. (Eds.: S. W. Baertschi, K. M. Alsante, R. A.
Reed), Informa Healthcare, London, 2011, pp. 10-48; d) B. M, R. D.
Patel, P. N. Prajapati, Y. K. Agrawal, J. Pharm. Anal. 2014, 4 (3), 159-
165.
[2]
[3]
a) L. K. Dow, M. M. Hansen, B. W. Pack, T.J. Page, S. W. Baertschi, J.
Pharm. Sci. 2013, 102 (5), 1404-1418; b) K. C. Waterman, AAPS
PharmSciTech 2011, 12 (3), 932-937.
a) X. Yan, R. M. Bain, R. G. Cooks, Angew. Chem. Int. Ed. 2016, 55,
12960–12972; Angew. Chem. 2016, 128, 13152–13166; b) X. Yan, H.
Cheng, R. N. Zare, Angew. Chem. Int. Ed. 2017, 56, 3562–3565;
Angew. Chem. 2017, 129, 3616–3619; c) C. Henry, C&EN, 2017, 95
(46), 16–18; d) C. Warschat, J. Riedel, Rev. Sci. Instrum. 2017, 88,
105108; (d) R. D. Davis, M. I. Jacobs, F. A. Houle, K. R. Wilson, Anal.
Chem. 2017. 89, 12494 – 12501.
Figure 4. Full MS (positive mode) showing (a) conversion of TFP (2) to 2a and
2b in bulk after reacting at 40 °C for 1523 min, achieving 50% conversion
ratio; (b) conversion of TFP (2) to 2a and 2b in ca. 30 µL Leidenfrost reactor
alllowed to shrink from 500 µL on the hotplate over ca. 90 s, achieving 68%
conversion ratio; (c) conversion of TCN (1) to 1a in bulk after reacting at 40 °C
for 1548 min, achieving 29% conversion ratio; (d) conversion of TCN (1) to 1a
in tiny (ca. 10 µL) Leidenfrost reactor allowed to shrink from 500 µL on the
hotplate for ca. 70 s, achieving 99% conversion ratio.
[4]
[5]
R. M. Bain, C. J. Pulliam, F. Thery, R. G. Cooks, Angew. Chem. Int. Ed.
2016, 55, 10478 –10482; Angew. Chem. 2016, 128, 10634 –10638.
a) C. W. Klampfl, M. Himmelsbach, Anal. Chim. Acta 2015, 890, 44-59;
b) G. A. Harris, L. Nyadong, F. M. Fernandez, Analyst 2008, 133,
1297–1301; c) G. A. Harris, A. S. Galhena, F. M. Fernández, Anal.
Chem. 2011, 83, 4508–4538; d) M. E. Monge, F. M. Fernández in
Ambient Ionization Mass Spectrometry (Eds.: M. Domin, R. Cody),
Royal Society of Chemistry, 2014, pp. 1-22; e) H. Chen, A. Wortmann,
W. Zhang, R. Zenobi, Angew. Chem. Int. Ed. 2007, 46, 580 –583;
Angew. Chem. 2007, 119, 586 –589; f) M.-Z. Huang, C.-H. Yuan, S.-C.
Cheng, Y.-T. Cho, J. Shiea, Annu. Rev. Anal. Chem. 2010, 3, 43–65; g)
Y. Wei, L. Chen, W. Zhou, K. Chingin, Y. Ouyang, T. Zhu, H. Wen, J.
Ding, J. Xu, H. Chen Sci. Rep. 2015, 5, 1–7; h) H. Li, P. Balan, A.
Vertes, Angew. Chem. Int. Ed. 2016, 55, 15035–15039; Angew. Chem.
2016, 128, 15259–15263; i) D. E. Damon, K. M. Davis, C. R. Moreira, P.
Capone, R. Cruttenden, A. K. Badu-Tawiah, Anal. Chem. 2016, 88,
1878−1884.
In conclusion, the reaction/analysis sequence of forced
degradation was achieved within minutes, in the three
degradation reactions studied. The nESI–MS and MS/MS
analysis show reproducible data when using Leidenfrost droplets
of constant volume for accelerated forced degradation and the
capability to rapidly assess the extent of degradation and
characterize degradant structure without using any separation
method. The experiments with droplets of different but constant
size also demonstrate an effect of surface/volume ratio on
reaction acceleration, and support the suggestion that interface
[6]
a) K. D. Schlecht, C. W. Frank, J. Pharm. Sci. 1975, 64 (2), 352-354; b)
B. Halling-Sørensen, G. Sengeløv, J. Tjørnelund, Arch. Environ.
Contam. Toxicol. 2002, 42, 263–271; c) J. A. Mollica, C. R. Rehm, J. B.
Smith, H. K. Govan, J. Pharm. Sci. 1971, 60 (9), 1380-1384; d) A. A.
Mahajan, A. K. Thaker, K. Mohanraj, J. Braz. Chem. Soc. 2012, 23 (3),
445-452; e) M. de Diego, J. Soto, S. Mennickent, J. Chromatogr. Sci.
2014, 52, 1193–1197; f) A. El-Gindy, B. El-Zeany, T. Awad, M. M.
Shabana, J. Pharm. Biomed. Anal. 2002, 27, 9–18. g) L. Min in Organic
Chemistry of Drug Degradation, RSC Publishing Cambridge, UK, 2012,
pp. 71-72.
is
a
significant factor. The levitated droplet shrinking
experiments suggested a quick, rough synthetic method. Future
prospects include further automation and multiplexing of the
reactions to make the method even more efficient. Given the
reliability and speed of the accelerated reaction/analysis and the
small quantity of API needed, this method is potentially valuable
in both drug discovery and development to (1) investigate the
stability of other APIs including biomolecules; (2) investigate the
stability of APIs in complex drug formulations, especially in
sterile solution formulations; and (3) speed up polymorph
screening.[8] The methodology developed here is potentially
applicable to accelerate processes in many other fields.
[7]
[8]
Z. Wei, M. Wleklinski, C. Ferreira, R. G. Cooks, Angew. Chem. Int. Ed.
2017, 56, 9386–9390; Angew. Chem. 2017, 129, 9514–9518.
R. Censi, P. Di Martino, Molecules 2015, 20, 18759-18776.
This article is protected by copyright. All rights reserved.

10.1002/chem.201801176
Chemistry - A European Journal
FULL PAPER
FULL PAPER
Traditional forced degradation of drug
substances and drug products can
take days to weeks but accelerated
reaction/analysis of active
Yangjie Li, Yong Liu, Hong Gao, Roy
Helmy, W. Peter Wuelfing, Christopher
J. Welch and R. Graham Cooks*
Page No. – Page No.
pharmaceutical ingredients was
achieved within minutes using
Accelerated Forced Degradation of
Pharmaceuticals in Levitated
Microdroplet Reactors
Leidenfrost droplets and MS analysis.
Lifetimes and sizes of the droplets and
degree of solvent evaporation control
reaction rate and product yield.
Interfacial effects increase reaction
rates in small droplets.
This article is protected by copyright. All rights reserved.