Investigation into the formation of the genotoxic impurity ethyl besylate in the final step manufacturing process of UK-369,003-26, a novel PDE5 inhibitor

Article abstract of DOI:10.1021/op100141g

Sulfonate esters have a demonstrated potential for genotoxicity, and therefore their potential presence at trace levels in active pharmaceutical ingredients (APIs) has recently raised concerns [Mesylate Ester Type Impurities Contained in Medicinal Product

Full text of DOI:10.1021/op100141g

Organic Process Research & Development 2010, 14, 1027–1031
Investigation into the Formation of the Genotoxic Impurity Ethyl Besylate in the
Final Step Manufacturing Process of UK-369,003-26, a Novel PDE5 Inhibitor
Yousef Hajikarimian,*,† Steve Yeo,‡ Robert W. Ryan,‡ Philip Levett,† Christopher Stoneley,† and Paul Singh†
Pfizer Global Research and DeVelopment, Chemical Research and DeVelopment, and Analytical Research and DeVelopment,
Ramsgate Road, Sandwich, Kent CT13 9NJ, United Kingdom
Abstract:
reaction between the excess benzenesulfonic acid and the ethoxy
side chain of the API. Sensitive and selective analytical methods
were also developed to detect and quantify subppm and higher
levels of ethyl besylate and deuterated analogues.
Sulfonate esters have a demonstrated potential for genotoxicity,
and therefore their potential presence at trace levels in active
pharmaceutical ingredients (APIs) has recently raised concerns
[Mesylate Ester Type Impurities Contained in Medicinal Products;
Swissmedic Department for Control of the Medicinal Products
Market, 23rd October 2007 and Hoog, T. J.-d. Request to Assess
the Risk of Occurrence of Contamination With Mesilate Esters and
Other Related Compounds in Pharmaceuticals; Coordination Group
for Mutual Recognition-Human Committee (CMDh), EMEA/
CMDh/98694/2008: London, 27 February, 2008]. Sulfonate salts
however, offer useful modification of physicochemical properties
of active pharmaceutical ingredients (APIs) containing basic
groups such that their use can at times offer significant advantages
over other counterions [Elder, D. P.; Delaney, E.; Teasdale, A.;
Eyley, S.; Reif, V. D.; Jacq, K.; Facchine, K. L.; Oestrich, R. S.;
Sandra, P.; David, F. The Utility of Sulfonate Salts in Drug
Development. J. Pharm. Sci. 2010, 99, 2948-2961; DOI: 10.1002/
jps.22058]. Indeed, the choice of benzenesulfonic acid as the
counterion for the UK-369,003 API afforded many advantages
over other salts such as citrate, hydrochloride, tartrate, and
phosphate as well as other sulfonate salts such as tosylate,
camsylate, and mesylate. The manufacturing route to the API
consists of two C-C bond-forming steps (steps 1 and 2/Scheme
1) and a final salt-formation step (step 3/Scheme 1). The step 2
cyclisation process involves the use of ethanol as the reaction
solvent. Residual levels of ethanol in the isolated product of the
step 2 process was initially thought to be responsible for the
formation of low levels of the genotoxic impurity ethyl besylate
(ppm levels) during the final step salt-formation process [Glow-
ienke, S.; Frieauff, W.; Allmendinger, T.; Martus, H. J.; Suter,
W.; Mueller, L. Mutat Res. 2005, 581, 23-34]. This was thought
to result from subsequent reaction of residual ethanol with
benzenesulfonic acid used in the final step (step 3). On the basis
of this mechanistic hypothesis, the levels of residual ethanol in the
isolated product from step 2 were controlled so that formation of
ethyl besylate would be minimised or avoided in the final step.
Spiking experiments coupled with deuterium labelling studies have
shed doubt on this mechanism of formation. Our experimental
results indicate that levels of ethyl besylate in the API are
independent of the level of residual ethanol in the step 2 product
(UK-369,003 free base) and are detected when higher than
stoichiometric amounts of benzenesulfonic acid are used in the
salt-formation process (step 3). This is thought to be due to a
1. Introduction
The use of sulfonic acids as suitable counterions for active
pharmaceutical ingredients (APIs) has increased significantly
over the past decade and now stands at 10% of the total cationic
counterions that are being developed. The increase in the usage
of such strong acids is at times ascribed to the general decrease
in the aqueous solubility of new drug candidates.5,6 In general,
APIs that exhibit low melting points often exhibit plastic
deformation during secondary processing which can cause both
caking and aggregation, thus impacting both flow and com-
pressibility performance.7 Sulfonate salts typically produce
higher-melting point versions of the API compared to those of
other salts of the API, which in addition to processing
advantages can enhance stability.8
UK-369,003 was nominated for development as the lead
candidate for treatment of benign prostatic hyperplasia (BPH).
The free base was found to be moderately crystalline with a
melting point of 168 °C. Solubility of the free base at
physiological pH was found to be poor hence necessitating a
(1) Mesylate Ester Type Impurities Contained in Medicinal Products;
Swissmedic Department for Control of the Medicinal Products Market,
23rd October 2007.
(2) Hoog, T. J.-d. Request to Assess the Risk of Occurrence of Contamina-
tion With Mesilate Esters and Other Related Compounds in Pharma-
ceuticals; Coordination Group for Mutual Recognition-Human Com-
mittee (CMDh), EMEA/CMDh/98694/2008: London, 27 February,
2008.
(3) Elder, D. P.; Delaney, E.; Teasdale, A.; Eyley, S.; Reif, V. D.; Jacq,
K.; Facchine, K. L.; Oestrich, R. S.; Sandra, P.; David, F. The Utility
of Sulfonate Salts in Drug Development. J. Pharm. Sci. 2010, 99,
2948-2961; DOI: 10.1002/jps.22058.
(4) Glowienke, S.; Frieauff, W.; Allmendinger, T.; Martus, H. J.; Suter,
W.; Mueller, L. Mutat Res. 2005, 581, 23–34.
(5) (a) Serajuddin, A. T. M. Salt Formation to Improve Drug Solubility.
AdV. Drug.DeliVery.ReV. 2007, 59, 603–616. (b) Elder, D. P. J. Pharm
Pharmacol. 2009, 61, 269–278.
(6) Serajuddin, A. T. M.; Pudipeddi, M. In Handbook of Pharmaceutical
Salts; Selection and Use (International Union of Pure and Aapplied
Chemistry; Stahl, P. H., Wermuth, G., Eds.; Wiley-VCH: NewYork,
2002; Chapter 6: Salt-Selection Strategies. pp 135-160.
(7) Lee, S.; Hoff, C. In Handbook Of Pharmaceutical Salts: Properties,
Selection and Use (International Union of Pure and Applied Chemistry;
Stahl, P. H., Wermuth, G., Eds.; Wiley-VCH: New York, 2002;
Chapter 8: Large Aspects of Salt Formation: Processing of Intermedi-
ates and Final Products. pp 191-220.
* Corresponding author. Telephone: +44 (1304) 642965. E-mail:
Yousef.Hajikarimian@pfizer.com.
(8) Elder, D. P. Physicochemical and Crystallographic Investigations into
the Salt Formation of Two Heterocyclic Drugs. Ph.D. Thesis;
University of Edinburgh: UK, 1992.
† Chemical Research and Development.
‡ Analytical Research and Development.
10.1021/op100141g  2010 American Chemical Society
Published on Web 06/25/2010
Vol. 14, No. 4, 2010 / Organic Process Research & Development
•
1027

Scheme 1. Manufacturing route to UK-369,003-26a
a CDI ) carbonyl diimidazole; MEK ) methyl ethyl ketone; EtOAc ) ethyl acetate; KOtBu ) potassium tertiary butoxide; EtOH ) ethanol.
comprehensive screen for a suitable salt form of the API.
Benzenesulfonic acid was found to form the most suitable
counterion for the API with a melting point of 248 °C and
satisfied all our requirements for primary and secondary
processing. The process for the formation of the benzenesulfonic
acid salt involved the use of water/methyl ethyl ketone (4%
water by volume) as the reaction medium. The water level at
4% ensured an optimum balance between product quality
(purging of impurities) and the reaction yield. The cyclisation
reaction (step 2/Scheme 1) involves the use of ethanol as the
reaction media. Any residual amount of ethanol in the isolated
step 2 product was therefore considered to be a considerable
risk factor in the potential formation of ethyl besylate during
the final step processing (step 3/Scheme 1).
The literature on the formation of sulfonate esters from
interaction between sulfonic acids and alcohols is sparse, whilst
there is much literature on the solvolytic instability of sulfonate
esters.9-13 Teasdale and co-workers demonstrated methanol
C-O bond cleavage of isotopically labelled methanol (18O-
label) in its reaction with methanesulfonic acid leading to the
formation of low levels of methyl methanesulfonate.14 This is
consistent with reversal of well established mechanisms for
solvolysis of sulfonate esters. Their studies of various reaction
profiles led to quantification of the levels of sulfonate ester
formed under conditions relevant to the formation of methane-
sulfonate salts of pharmaceutically active bases. Their work
demonstrated a clear scope to select conditions for the prepara-
tion of sulfonate salts in alcoholic solvents to minimize
formation of sulfonate esters. Incorporation of water into the
reaction system as well as eliminating excesses of sulfonic acid
used are considered vital in avoiding formation of the sulfonate
ester.14
Water was indeed incorporated in the process for the
manufacture of the API (step 3/Scheme 1) and charge of the
benzenesulfonic acid was not allowed to exceed the stoichio-
metric level. Subsequent to development of a reliable methodol-
ogy for determination of residual ethyl besylate in isolated API
(UK-369,003-26), 22 batches of API were tested for ethyl
besylate. All batches showed levels <1 ppm. The level of
residual ethanol in the free base (UK-369,003) leading to these
batches of API was evaluated to be no more than 200 ppm.
Our assumption was therefore such that a batch of free base
UK-369,003 containing <200 ppm ethanol would lead to API
with <1 ppm ethyl besylate. A maximum limit of 1 ppm ethyl
besylate in API was considered to be our desired internal
(9) Winstein, S.; Grunwald, E.; Jones, H. W. J. Am. Chem. Soc. 1951,
73, 2700.
(10) Robertson, R. E. Can. J. Chem. 1953, 31, 589.
(11) Smith, M. B.; March, J. March’s AdVanced Organic Chemistry; Wiley:
New York, 2001; p 464.
(12) Isaacs, N. Physical Organic Chemistry; Prentice Hall: Harlow, 1995;
p 418.
(13) Bentley, T. W.; Bowen, C. T.; Brown, H. C.; Chloupek, F. J. Org.
Chem. 1981, 46, 38.
(14) Teasdale, A.; Eyley, S. C.; Delaney, E.; Jacq, K.; Taylor-Worth, K.;
Lipczynski, A.; Reif, V.; Elder, D. P.; Facchine, K. L.; Golec, S.;
Oestrich, R. S.; Sandra, P.; David, F. Org. Process Res. DeV. 2009,
13, 429–433.
1028
•
Vol. 14, No. 4, 2010 / Organic Process Research & Development

specification for this genotoxin. It is also worth noting that a
‘threshold of toxicological concern’ (TTC) approach which is
stated by the European Medicines Agency (EMEA) would limit
the daily patient exposure to no more than (NMT) 1.5 µg/day
for supporting a market application (i.e., 15 ppm).15
Higher than 1 ppm ethyl besylate was detected in some of
our small-scale laboratory batches as well as one pilot-plant
batch of API. In studying the impact of using excess benze-
nesulfonic acid (>1 equiv) in the step 3 reaction, it soon became
evident that this leads to API with >1 ppm ethyl besylate. More
noteworthy was the fact that the in-going free base UK-369,003
did not contain residual ethanol at >200 ppm.
Further studies to investigate the cause for the formation of
ethyl besylate began to shed doubt on the rationale of our initial
hypothesis. Our findings led us to focus on the stoichiometry
of the benzenesulfonic acid used in the step 3 salt-formation
process.
2. Experimental Section
2.1. General Procedure for the Synthesis of API (UK-
369,003-26). To a reaction vessel was charged free base UK-
369,003 (50 g, 0.096 mol). To this was added a mixture of
methyl ethyl ketone (480 mL, 9.6 mL/g) and water (13.15 mL,
0.26 mL/g). The reaction mixture was heated to 70 °C over a
period of 30 min to achieve dissolution. Benzenesulfonic acid
solution (70% (w/w); 21.75 g, 0.1 mol) was added to the hot
solution in four equal portions at hourly intervals. Product
crystallisation was observed at the end of the first hour. The
reaction was cooled to 0 °C at a rate of 0.1 °C/min. It was then
further granulated at 0 °C for a period of 3 h. The white solid
product was collected by filtration, and the wet cake was washed
with methyl ethyl ketone (4 mL/g). The solid product was dried
in a vacuum (50 °C) for a period of 12 h. The yield was in the
order of 96%.
2.2. Analytical Method Development for Trace Levels
of Ethyl Besylate. Due to a batch-dependent matrix effect,
which was observed as variable recovery relative to external
standards, the method utilized a standard addition approach. This
was applied either as a dual level (zero and 1 ppm) limit test or
multiple level (zero, 0.5 ppm, 1 ppm, and 2 ppm) addition for
the purpose of validation.
Samples were dissolved in 0.1 M sodium hydroxide at 10
mg/mL; 0.5 mL of hexane or hexane containing ethyl besylate
at appropriate spiking levels was added. The mixture was shaken
and allowed to separate, and the upper organic layer was
recovered for injection into a GC/MS. Experiments were
performed, introducing the ethyl besylate spike as a methanolic
solution into the aqueous test. These experiments showed that
exposure of the spike to the test prior to extraction was not a
significant factor in the determination. The ethyl besylate spike
was therefore introduced in the extraction solvent for ease of
use.
The challenges presented in developing trace analytical
methodology have been well documented. In particular matrix
effects can have significant impact on the ability of the analyst
to provide accurate and precise results robustly. These matrix
effects can raise issues around recovery, precision, contamina-
tion, interference, and the legitimacy of results. To reduce these
potential matrix effects it is necessary to remove the matrix at
a stage in the analysis before the effects become active.
UK-369003 presented such challenges from the outset of
method development where a simple, large-volume injection
(LVI) gas chromatographic analysis with mass spectroscopic
detection (GC/MS) was employed. Although the analysis
appeared to proceed normally, positive results were obtained
far in excess of those possible from the residual ethanol present
in the batch. It was theorised that the matrix (API or impurity
other than ethanol), was producing the ethyl besylate in the GC
inlet during the desolvation and volatilisation process. A method
developed in our laboratory that is generic for alkylators16 and
excludes the matrix from the chromatographic system was
employed. The method utilises pentafluorothiophenol as a
derivatising agent which is readily alkylated into a headspace
(HS) and GC amenable species, with enhanced MS detectability.
Although this form of sample pretreatment goes some way to
reduce matrix effects, the test sample is heated in the HS oven
with caustic and reactive reagents and cannot therefore be called
universally applicable. UK-369003-26 produced positive results
with this method, and a milder extraction procedure was sought.
Liquid-liquid extraction has been well utilised in trace analysis
and is still widely applied despite more automatable alternatives
being available, e.g. solid-phase extraction (SPE), solid-phase
micro-extraction (SPME), and stir-bar sorptive extraction
(SBSE). It can provide a quick solution to method development
where there is sufficient aqueous solubility and affinity of the
matrix with sufficient lipophilicity of the target. It can also be
a relatively quick technique to employ as equilibration/extraction
times can be short relative to SPME or SBSE.
GC Conditions. Agilent 6890 GCs with 5975 or 5973 MSD
detectors were used. Thermo-Fisher Trace GC and DSQ II MS
were also used.
Two columns were employed which gave acceptable chro-
matographic performance. Due to the selectivity of the extrac-
tion and detection no specificity issues were encountered with
either configuration.
Column 1 was a VF-5MS (Varian Inc., model CP912092),
20 m × 0.1 mm i.d. × 0.2 µm film thickness.
(15) Guideline on the Limits of Genotoxic Impurities, CPMP/SWP/5199/
02, EMEA/CHMP/QWP/251344/2006; Committee for Medicinal
Products (CHMP), European Medicines Agency (EMEA): London,
28 June 2006.
Column 2 ZB-5MS (Phenomenex Ltd.), 20 m × 0.18 mm
× 0.18 µm film.
The oven temperature program for column 1 was 40 °C held
for 2 min, 30 °C/min to 300 °C and held for 1.33 min. PTV
(16) Alzaga, R.; Ryan, R. W.; Taylor-Worth, K.; Lipczynski, A.; Szucs,
R.; Sandra, P. J. Pharm. Biomed. Anal. 2007, 45, 472–479.
Vol. 14, No. 4, 2010 / Organic Process Research & Development
•
1029

Scheme 2a
injection at 30 °C held for 1 min then ramped to 250 °C at 500
°C/min. Injection volume was 5 µL, and the split flow was 100
mL/min for 1 min, splitless during the temperature ramp, and
30 mL/min after 2 min. SIM detection was used monitoring
m/z 77, 141, and 186.
Column 2 used a split/splitless injector in splitless mode (50
mL/min split at 1.0 min), with a 1 µL injection. The inlet was
maintained at 250 °C. The oven temperature program was 60
°C for 0.5 min then 40 °C/min to 250 °C. Column flow was
1.5 mL/min constant.
a EtB ) ethyl besylate; MEK ) methyl ethyl ketone.
2.3. Analytical Performance. In the single step, extraction
employed 75% of ethyl besylate partitioned into the organic
phase. Limits of detection were ,1 ppm. Injection repeatability
(n ) 6) typically gave less than 10% relative standard deviation
(RSD), at the 1 ppm relative level. Repeatability in the presence
of the test (n ) 3, independent extractions) gave an RSD of
2%. Linearity with extracted external standards over the range
0.1-2 ppm gave a coefficient of determination r2 of 99.9%.
Linearity of ethyl besylate response over the spiking range 0-2
ppm (in a test containing 2 ppm ethyl besylate) gave an r2 of
98.9%. Recovery against extracted standards was variable and
between 15 and 120%, but as previously indicated this was
compensated for by the application of a standard addition
approach to the determination.
Table 1. Step 3 salt-formation process
ethanol spike
(ppm)
ethyl besylate
level in API (ppm)
experiment
control
A
B
111
500
10000
0.18
0.06
0.08
Table 2. Step 3 process using an overcharge of BSA (2
a
equiv) and spiked with ethanol-d6
experiment
number
ethyl besylate
(ppm) in API
ethyl besylate-d5
(ppm) in API
A
1671
ND
a ND ) none detected; BSA ) benzenesulfonic acid.
In order to understand the significance of the results,
repeatability was performed for the analytical procedure and
the experimental procedure including batch synthesis. Analytical
precision gave RSDs of 1.4% (n ) 6) for injection, and 15.7%
(n ) 3) for multiple test preparations and extractions. Experi-
mental precision was obtained from the results of the determi-
nation of ethyly besylate in three batches of UK-369003
synthesised in parallel, using identical conditions and reagents.
An RSD of 55% indicated that the variability present in
chemical synthesis, which affected the actual ethyl besylate
content of UK-369003 batches, was more significant than the
analytical variability.
amount of ethyl besylate is formed. This is in line with the
findings reported by Teasdale and co-workers when low levels
of methyl methanesulfonate were formed from interaction
between methanol and methanesulfonic acid.14
In order to provide further supporting evidence that the
presence of residual ethanol in the free base UK-369,003 is
the main contributor towards the formation of the ethyl besylate
during the step 3 salt-formation process, we carried out a series
of spiking experiments.
Our current salt-formation process (step 3) was spiked with
varying levels of ethanol (500 and 10,000 ppm), and the isolated
API was tested for residual levels of ethyl besylate and
compared with that of the control experiment. Our results
indicated no obvious correlation between residual ethanol in
UK-369,003 free base and ethyl besylate in the isolated API
(Table 1).
This finding placed doubt on the validity of our initial
mechanistic hypothesis for the ethyl besylate formation.
We also investigated the effect of increasing the stoichiom-
etry of the benzenesulfonic acid used in the salt-formation
process whilst keeping other factors unchanged. The input UK-
369,003 free base contained no residual ethanol. Our results
indicated that an increase in the stoichiometry of benzene-
sulfonic acid (0.2 equiv) led to an increase (>1 ppm) in the
residual level of detected ethyl besylate in the isolated API.
We needed to understand the origin of the ethyl moiety of
the resultant ethyl besylate in the API.
Overall the validation data demonstrated that the analytical
procedure was suitable for the accurate and precise determi-
nation of ethyl besylate in UK-369003 but that care was required
to compare batch results and to draw significance from trends.
Results and Discussion
Our initial mechanistic hypothesis that drove our plans for
limiting the residual level of the ethyl besylate in the API was
based on controlling the level of residual ethanol in the isolated
free base UK-369,003. It was thought that a reaction between
ethanol and benzenesulfonic acid during the course of the salt-
formation process (step 3/Scheme 1) would lead to the forma-
tion of ethyl besylate.
Our laboratory experimentation (Scheme 2) did indeed show
formation of low levels of ethyl besylate (4-6 ppm) over a
period of 6 h and increased levels (>1000 ppm) over prolonged
reaction time when ethanol (100 mol equiv) and benzenesulfonic
acid (1 mol equiv) were refluxed in methyl ethyl ketone in the
presence of low levels of water (∼4% by volume). (The reaction
condition used here was to mimic the step 3 salt-formation
process.). Our data indicate that even in the presence of a large
excess of ethanol under reflux conditions, only a relatively small
If residual ethanol in the input UK-369003 free base plays
no part in the formation of the ethyl besylate, the ingoing free
base itself must therefore be involved in the formation of this
impurity. In order to test this hypothesis a salt-formation process
(step 3), with an overcharge of the acid spiked with ethanol-d6,
was carried out. It resulted in API with no detectable level of
1030
•
Vol. 14, No. 4, 2010 / Organic Process Research & Development

Table 3
experiment
number
ethyl besylate (ppm)
detected in API
ethyl besylate-d5
(ppm) detected in API
A
10
1100
ethyl besylate-d5 but a significant level of ethyl besylate (1671
ppm) (Table 2).
Analysis of the isolated API showed the presence of a
significant level of ethyl besylate-d5 (1100 ppm) and a very
low level of ethyl besylate (10 ppm) (1H NMR of the ingoing
deuterated UK-369003 showed the presence of a residual
amount of the non-deuterated UK-369003. This in turn explains
the presence of ethyl besylate at 10 ppm.)
In view of this result, we developed the following hypothesis
as an explanation for the origin of the ethyl besylate. The in-
going free base UK-369003 can react with any excess (>1
equiv) benzenesulfonic acid, leading to the formation of the
undesired ethyl besylate.
This finding is consistent with our revised mechanism for
the formation of ethyl besylate in the final step 3 salt-formation
We therefore postulated the following mechanism:
Our results indicated that a slight excess of BSA (>1 equiv)
charge could lead to the formation of undesirable levels of the
ethyl besylate during the final step salt formation. It has to be
noted that residual levels of pyridone (also formed in step 2
reaction and carried forward through to the step 3) are normally
detected at slightly higher levels in API, and we consider this
increase to be partly associated with our proposed mechanism
shown above.
In an attempt to further verify our hypothesis, a
deuterated analogue of UK-369003 (i.e., protons of the
ethoxy side chain were substituted by deuterium) was
subjected to the step 3 reaction process in the presence
of excess acid (2 equiv) as well as residual ethanol (10,000
ppm) (Table 3).
process.
Acknowledgment
We thank Dr. Ron Ogilvie and Dr. Andrew Lipczynski
(Pfizer Global R&D) for their valuable advice and discussions
during the preparation of this manuscript. We also thank Dr.
Laurence Harris of Pfizer Chemical R&D department as well
as Dr. Terry Kougoulos of Material Sciences for their contribu-
tions to this project.
Received for review May 21, 2010.
OP100141G
Vol. 14, No. 4, 2010 / Organic Process Research & Development
•
1031