Process development toward the pilot scale synthesis of the piperidine-based cocaine analogue and potent dopamine and norepinephrine reuptake inhibitor CTDP 31,446

Article abstract of DOI:10.1021/op060114g

(+)-Methyl 4β-(4-chlorophenyl)-1-methylpiperidine-3α-carboxylate hydrochloride (CTDP 31,446) is known as a dopamine reuptake inhibitor. This cocaine analogue lacking the tropane skeleton is being considered for potential treatment of cocaine addiction. Herein we report the development of a scalable process for the preparation of this compound. This study was mainly aimed at improving the process throughput, eliminating chromatographic purifications for the separation of (±)-6-cis isomer from (±)-6-trans isomer, and developing a robust crystallization for isolation of pure (±)-6-cis in a single crop with a good mass recovery. The process development work also highlights an efficient recycle of (±)-6-trans via a kinetic epimerization followed by crystallization of the resulting (±)-6-cis isomer. The resolution of (±)-6-cis and the crystallization of the final HCl salt were optimized and implemented to afford CTDP-31,446 with high purity and good mass recovery.

Full text of DOI:10.1021/op060114g

Organic Process Research & Development 2006, 10, 914−920
Process Development toward the Pilot Scale Synthesis of the Piperidine-Based
Cocaine Analogue and Potent Dopamine and Norepinephrine Reuptake Inhibitor
CTDP 31,446
Bingidimi I. Mobele,* Taryn Kinahan, Luckner G. Ulysse, Steven V. Gagnier, Michael D. Ironside, Graham S. Knox, and
Farahnaz Mohammadi
Chemical DeVelopment Department, Albany Molecular Research Inc., 21 Corporate Circle, P.O. Box 15098,
Albany, New York 12212, U.S.A.
Abstract:
+)-Methyl 4â-(4-chlorophenyl)-1-methylpiperidine-3r-carbox-
tropine analogues, methylphenidate analogues, mazindol
analogues, piperazine analogues (also called GBR series, 4),
and most recently piperidine analogues (3). Most notably in
the latter category, piperidine 3-carboxylic acid esters bearing
chlorophenyl substituents in position 4 have been extensively
studied as possible dopamine reuptake inhibitors. This series
of compounds, which may be viewed as truncated versions
of the WIN series, was first reported by Clarke and
(
ylate hydrochloride (CTDP 31,446) is known as a dopamine
reuptake inhibitor. This cocaine analogue lacking the tropane
skeleton is being considered for potential treatment of cocaine
addiction. Herein we report the development of a scalable
process for the preparation of this compound. This study was
mainly aimed at improving the process throughput, eliminating
chromatographic purifications for the separation of (()-6-cis
isomer from (()-6-trans isomer, and developing a robust
crystallization for isolation of pure (()-6-cis in a single crop
with a good mass recovery. The process development work also
highlights an efficient recycle of (()-6-trans Wia a kinetic
epimerization followed by crystallization of the resulting (()-
3
co-workers in 1973 and has received revived interest more
recently, most notably by Kozikowski and co-workers.2
To prepare kilogram quantities of CTDP-31,446 needed
for clinical trials, a scalable process was required. The
existing synthetic route provided by the National Institute
on Drug Abuse (NIDA) is based on work by Kozikowski et
a,4
4
6-cis isomer. The resolution of (()-6-cis and the crystallization
al., who prepared a large number of potent piperidine-based
of the final HCl salt were optimized and implemented to afford
CTDP-31,446 with high purity and good mass recovery.
analogues. This route involved initial Grignard addition of
4-chlorophenylmagnesium bromide to arecoline to generate
a mixture of racemic (()-6-cis and (()-6-trans (Scheme 1).
The (()-6-cis was crystallized from ethyl acetate/hexanes
and recovered in 22% yield in a first crop. Further recovery
from the mother liquors (34%) was accomplished Via
chromatography, which also yielded 18% of the (()-6-trans
isomer. To prepare CTDP-31,446, the authors took advantage
of the relative ease of resolution of (()-6-cis to generate
enantiomerically homogeneous (-)-6-cis, which was then
epimerized to the desired (+)-6-trans isomer with sodium
methoxide in refluxing methanol. Subsequent treatment of
the enantiomerically pure (+)-6-trans with HCl (g) in
methanol then afforded CTDP-31,446. Noteworthy and
critical to the choice of this approach was the difficulty in
achieving a direct resolution of (()-6-trans.
Introduction
Cocaine (1) is a powerful, abused drug with strong
reinforcing properties. Several investigations have established
that cocaine binds to neuronal proteins including the dopam-
ine transporter (DAT), the serotonin transporter (SERT), and
the norepinephrine transporter (NET), resulting in its rein-
1
forcing effect. It is now well-established that the binding
of cocaine to the DAT results in the inhibition of the reuptake
2
of DA after synaptic transmission in the brain. The
development of drugs for the treatment of cocaine addiction
has been a main focus of several research laboratories
worldwide. These extensive investigations have resulted in
the development of different classes of molecules targeting
the DAT with the aim of modulating the action of cocaine.
These DAT inhibitors are categorized mainly into tropane
analogues (mostly compounds of the WIN series, 2), benz-
For this route to be efficiently and safely scaled up, a
number of significant process issues needed to be addressed.
4
(
1) The original procedure employed large amounts (∼90
volumes) of diethyl ether as the reaction solvent, which
would significantly impact the process throughput on scale.
In addition, the inherent hazards posed by the use of the
highly flammable diethyl ether on scale were a concern. (2)
Serious stirring issues stemmed from the generation of a thick
gel during the Grignard reaction, particularly when carried
out under less dilute conditions, making an inverse quench
*
Author for correspondence. E-mail: Bingidimi.Mobele@albmolecular.com.
Telephone: +1 518 464 0279. Fax: +1 518 464 0289.
(
1) For general reviews, see: (a) Smith, M. P.; Hoepping, A.; Johnson, K. M.;
Trzcinska, M.; Kozikowski, A. P. Drug DiscoVery Today 1990, 4, 322. (b)
Howell, L. L.; Wilcox, K. M. J. Pharmacol. Exp. Ther. 2001, 298 (1), 1.
(c) Carroll, F. I.; Lewin, A. H.; Kuhar, M. J. Neurotransm. Transp. 1997,
2
63. (d) Kuhar, M. J.; Ritz, M. C.; Boja, J. W. Trends Neurosci. 1991, 14
(
7), 299. (e) Singh, S. Chem. ReV. 2000, 100, 925.
(3) Clarke, R. L.; Daum, S. L.; Gambino, A. J.; Aceto, M. D.; Pearl, J.; Levitt,
M.; Cumiskey, W. R.; Bogado, E. F. J. Med. Chem. 1973, 16 (1), 1260.
(4) Kozikowski, A. P.; Araldi, G. L.; Boja, J.; Meil, W. M.; Johnson, K. M.;
Flippen-Anderson, J. L.; George, C.; Saiah, E. J. Med. Chem. 1998, 41,
1962.
(
2) (a) Tamiz, A. P.; Zhang, J.; Flippen-Anderson, J.; Zhang, M.; Johnson, K.
M.; Deschaux, O.; Tella, S.; Kozikowski, A. P. J. Med. Chem. 2000, 43,
1
215. (b) Carroll, F. I.; Pawlush, N.; Kuhar, M. J.; Pollard, G. T.; Howard,
J. L. J. Med. Chem. 2004, 47, 296.
9
14
•
Vol. 10, No. 5, 2006 / Organic Process Research & Development
10.1021/op060114g CCC: $33.50 © 2006 American Chemical Society
Published on Web 08/19/2006

Figure 1.
Scheme 1
purification prior to conversion to the corresponding HCl
salt.
Optimization of the Grignard Addition. It is interesting
to note that the addition of various Grignard reagents to
6
arecoline as originally described by Plati et al. remarkably
generates predominantly the adducts derived from a Michael
addition rather than products from reaction at the ester moiety
when carried out in diethyl ether. Attempts to use other
ethereal solvents such as THF, typically employed in
Grignard addition reactions, provide little or none of the
desired 1,4-conjugate addition products. Instead, adducts
derived from the competing 1,2-addition to the ester
5
b
predominate.
Initial attempts to avoid the use of diethyl ether altogether
by using THF in the presence of catalytic amounts of copper
additives (up to 10 mol %), such as CuI or Li Cu Cl , were
2 2 4
unsuccessful. In all cases, a faster addition to the ester was
prevalent, leading to a presumed intermediate R,â-unsaturated
ketone which then underwent a further 1,4-addition resulting
in an overall incorporation of two 4-chlorophenyl units. It
is worth noting that successful Grignard addition reactions
to related R,â-unsaturated esters have been accomplished in
the presence of up to 50 mol % of a CuBr‚Me
2
S complex in
tetrahydrofuran as the reaction solvent. For example, Casamit-
7
a
jana and co-workers showed that the conjugate addition of
ethylmagnesium bromide or phenylmagnesium bromide to
3
-benzyloxycarbonyl-5,6-dihydropyridin-2-one afforded 1,4-
adducts in good yields in tetrahydrofuran using 50 mol %
of the CuBr‚Me S complex. Similarly, Davies and co-
workers have also successfully employed 50 mol % of the
CuBr‚Me S complex in tetrahydrofuran at 0 °C to ambient
temperatures to achieve the conjugate addition of Grignard
reagents to anhydroecgonine derivatives and related R,â-
unsaturated esters incorporating a bridging heteroatom in the
bicyclic framework, such as 6-azabicyclo[3.2.2]non-3-ene
and 6-azabicyclo[3.2.2]non-2-ene derivatives. In our case,
2
7
b
such as that employed in the original procedure difficult to
implement on scale. It should be noted that this stirring issue
has been previously observed and documented by others in
2
5
related Grignard addition reactions to arecoline. (3) Isolation
of the desired (()-6-cis isomer was not straightforward and
necessitated chromatography of the mother liquors from the
initial low-yielding crystallization. (4) Additionally (()-6-
trans was isolated as an oil which required chromatographic
(
(
6) Plati, J. T.; Ingerman, A. K.; Wenner, W. J. Org. Chem. 1957, 22, 261.
7) (a) Ecija, M.; Diez, A.; Rubiralta, M.; Casamitjana, N.; Kogan, M. J.; Giralt,
E. J. Org. Chem. 2003, 68, 9541. (b) Davies, H. M. L; Hodges, L. M.;
Gregg, T. M. J. Org. Chem. 2001, 66, 7898.
(5) (a) Crowe, D.; Ward, N.; Wells, A. S. World Patent WO 01/29032. (b)
Ward, N.; West, V. World Patent WO 02/32870.
Vol. 10, No. 5, 2006 / Organic Process Research & Development
•
915

Table 1. Dependence of Isomer Ratio on Quench Method
normal quench at various temperatures by
inverse quench of
adding TFA to bulk reaction mixturesb
in-process aliquots
a
reaction
temp (°C)
into aq. NH
4
Cl
quench
quench
time
cis:trans
entry
cis:trans ratio
temp (°C)
ratio
1
2
3
4
-40 to -30
-30 to -25
-20 to -10
-10 to -5
not measured
2.8:1
3:1
-75 to -70
-72 to -60
-15 to 12
2 h
1:3.8
1.6:1
1:1
25 min
6 min
2 min
3:1
-42 to -30
1:1
a
In-process check by inverse quench of small aliquots into excess saturated aqueous NH
4
Cl prior to GC analysis. b Quench upon completion of reaction.
the use of the CuBr‚Me
mentally unacceptable on scale.
2
S complex was deemed environ-
of (()-6-cis afforded variable ratios of the stereoisomers,
generally favoring the thermodynamic (()-6-trans adduct.
A careful examination of the isomeric distribution de-
pendence on the quench method revealed that the amount
of kinetic (()-6-cis product was significantly increased when
the quench was carried out fast, regardless of the reaction
temperature at which the Grignard addition had been
conducted (Table 1). Particularly noteworthy, and perhaps
more significant, was the observation that when small
aliquots were withdrawn from the reaction mixtures and
inversely quenched into saturated aqueous ammonium chlo-
ride, the (()-6-cis isomer was consistently and reproducibly
generated as the major component (cis/trans ratio ) 2.8:1
to 3:1; see entries 2-4 in Table 1). These experiments clearly
established that the inverse quench method provided es-
sentially an instantaneous protonation of the enolate and
could, therefore, be used advantageously for improving the
isomeric ratio of the Grignard addition product.
Therefore, given the tight timelines, no further evaluation
of the conjugate addition reaction in alternative solvents in
the presence of other copper additives was pursued, and
efforts were focused instead on optimizing the Grignard
addition reaction in diethyl ether with the following objec-
tives: (1) to reduce the amount of diethyl ether by using an
appropriate cosolvent; (2) to improve the process throughput
by reducing the amount of solvent without compromising
the stirring efficiency; (3) to maximize the cis-selectivity in
the Grignard addition reaction and optimize product recovery
by crystallization; and (4) to further improve the mass
recovery of the desired (()-6-cis product by recycling (()-
6
-trans.
The decision to focus on the isolation and resolution of
-)-6-cis was based on two considerations: (a) unlike the
(
trans isomer, which was an oil at ambient temperature and
difficult to purify in a single step Via nonchromatographic
methods, the cis isomer was easily crystallized from
The predominant formation of (()-6-cis upon rapid
quench has not been fully investigated and is not completely
understood. Nonetheless, the experimental results can be
rationalized by considering that a potentially favored chelated
kinetic enolate depicted by structure 7 (Figure 2) is pre-
dominantly generated upon conjugate addition of the Grig-
nard reagent. A fast protonation of 7 in the equatorial
direction, which corresponds to the most accessible face
of the π-system, would provide (()-6-cis as the major
8
heptanes or heptanes/dichloromethane, heptanes/ethyl acetate
and could be isolated with high purity; (b) the cis isomer
was readily resolved with good mass recovery, whereas the
resolution of the trans isomer was problematic and would
have required a substantial development and optimization
9
time.
Initial trial reactions were conducted by slowly adding a
solution of arecoline in 5 volumes of ether to a cold Grignard
solution (1 M solution in ether, 2 equiv, and ∼9 volumes)
at -30 °C. In-process GC analysis revealed that the reaction
was essentially complete within an hour after completing
the Grignard addition. Furthermore, careful examination of
the reaction profile indicated that only 1.4 equiv of Grignard
reagent were typically required to achieve complete con-
sumption of arecoline. Subsequent treatment of the reaction
mixture at low temperature (-70 °C) with TFA in the hopes
of quenching the kinetic enolate to maximize the recovery
1
0
isomer.
Conversely, when a comparatively slower quench is
conducted, the initially generated (()-6-cis may undergo in
situ epimerization to afford the thermodynamic adduct (()-
6-trans. The limited experimental data gathered in the course
of this study do not provide enough evidence to suggest to
(
9) At the inception of the process development work and concurrently with
the optimization of the Grignard reaction, several resolving agents were
evaluated for their ability to provide a direct resolution of (()-6-trans. These
included (-)-dibenzoyl-L-tartaric acid, (S)-(+)-mandelic acid, L-(+)-tartaric
acid, di-p-toluoyl-L-tartaric acid, D-(+)-malic acid, dipivaloyl-L-tartaric acid,
(S)-(-)-2-pyrrolidone-5-carboxylic acid, (1S)-(+)-10-camphorsulfonic acid,
(
8) The isolation and purification of (()-6-trans Via crystallization entailed the
following chemical and process steps: (a) treatment of the crude extracts
from the Grignard reaction with sodium methoxide after solvent exchange
into methanol to maximize the amount of (()-6-trans upon thermodynamic
epimerization of the cis isomer, (b) conversion of crude (()-6-trans into
the corresponding HCl salt in methanol and crystallization, and (c)
regeneration of the free base by treatment of the salt with aqueous base
and extraction. Only a modest 33% yield was achieved under these
conditions. Overall, the number of steps required for generating material
of suitable purity for a direct resolution of (+)-6-trans presented no obvious
throughput advantage compared to the isolation and elaboration of the (()-
and D-(-)-quinic acid. Most of these resolving agents provided low to no
enantiomeric enhancement in pilot screening experiments in a variety of
solvents, to the exception of di-p-toluoyl-L-tartaric acid, which afforded 31%
ee and 29% ee in methyl isobutyl ketone (MIBK) and ethyl acetate,
respectively, using 1 molar equiv of the resolving agent. The chiral purity
could be increased up to 71% ee using substoichiometric amounts of di-
p-toluoyl-L-tartaric acid, albeit with a lower mass recovery. Due to time
constraints, however, further optimization of the direct resolution of (()-
6-trans was not pursued, and efforts focused instead on streamlining the
resolution of (()-6-cis.
(10) For a review on the stereochemistry of kinetic protonation of enolates in
6
-cis isomer.
cyclic systems, see: Zimmerman, H. E. Acc. Chem. Res. 1987, 20, 263.
916
•
Vol. 10, No. 5, 2006 / Organic Process Research & Development

exchange into heptanes and a cooling crystallization were
implemented, which allowed the isolation of (()-6-cis
consistently with 92-97% isomeric purity, as determined
by GC analysis, and with good mass recovery.11 The
crystallization was initially carried out on a rotary evaporator
(entries 1 and 2, Table 2) but was subsequently optimized
and demonstrated in the third batch by carrying out a formal
distillation to generate a supersaturated solution. The batch
was then naturally cooled to ambient temperature, and the
crystallization was allowed to proceed for 16 h to maximize
the mass recovery in a single crop (entry, Table 2).
An additional amount of (()-6-cis was readily obtained
by carrying out a kinetic epimerization of (()-6-trans from
the mother liquors (cis/trans ratio in mother liquors for the
Figure 2.
Table 2. Grignard Reaction Scale-Up Runs Using Inverse
Quench into 2 M aq HCl
3
50 g scale batch, 1:6) using LDA as the base and THF as
the solvent, then implementing an inverse quench into
aqueous HCl and a selective crystallization of (()-6-cis (Vide
supra; cis/trans ratio of isolated product, 8:1 for the 350 g
scale batch). By applying this recovery protocol, a combined
mass recovery of 75-80% was achieved on a 200-350 g
scale, with a single recycle of (()-6-trans.
arecoline cis/trans ratio
% yield
entry
(g)
crude product crystallization
remarks
single cropa
1
2
3
50
200
350
2.7:1
2.6:1
2.5:1
50
54
53
two cropsb
c
single crop
Resolution of (()-6-cis with (-)-Dibenzoyl-L-tartaric
Acid. With an efficient protocol to generate large amounts
of (()-6-cis in place, efforts were focused on optimizing
the resolution procedure and streamlining the generation and
isolation of the final HCl salt. Previous investigations into
the resolution procedure showed that when (()-6-cis was
treated with 1 mol equiv of the resolving agent in 12 volumes
of methanol at ambient temperature overnight, the desired
a
Cis/trans ratio of isolated product, 12:1. b Cis/trans ratio of first crop, 39:1.
c
Cis/trans ratio of isolated product, 12:1
what extent protonation of the enolate from the axial direction
is competitive with epimerization of (()-6-cis to account
for the formation of (()-6-trans. Additional future investiga-
tions using for instance in situ IR spectroscopy may shed
some light on the mechanistic details of this reaction.
Upon incremental scale-up, minor variations in the
isomeric distribution were observed that may somewhat
reflect mass transfer efficiency differences at the quench
(-)-6-cis was obtained with 90% enantiomeric excess and
35-40% mass recovery. The chiral purity could be improved
to 94-98% ee by either working under more dilute condi-
tions (20-30 volumes of methanol) or using substoichio-
metric amounts of the resolving agent.
(
Table 2, inverse quench into 2.0 M aqueous HCl). However,
a systematic evaluation of processing parameters likely to
impact the mass transfer, such as mixing efficiency at the
quench and its impact on the isomeric distribution, was not
pursued during this study.
It is worth pointing out that methyl tert-butyl ether
MTBE) was found to be a convenient and safe cosolvent
In an effort to further improve the resolution procedure
with a view to generating (-)-6 with g98% ee without a
need for a recrystallization, the resolution was further studied
at a slightly higher temperature (30 °C). It was expected that,
by implementing the isolation at this temperature, most of
the more soluble, unwanted diastereomeric salt would remain
in solution, thus making possible the recovery of the desired
diastereoisomer with higher chiral purity. The resolution was
thus carried out using 1 mol equiv of the resolving agent
and 15 volumes of methanol, by adding a slurry of (()-6-
cis in methanol to a warm solution (50-60 °C) of (-)-
dibenzoyl-L-tartaric acid in methanol. The mixture was
briefly heated at reflux and then slowly cooled to 30 °C over
(
compatible with the Michael addition reaction, and it was
therefore used in all runs reported in Table 2. This allowed
us to further reduce the amount of diethyl ether in the process
from 17 to 7 volumes, by restricting its use to the amount
present in the commercial Grignard solution. Under the
optimized conditions, 5.0-5.5 volumes of MTBE cosolvent
were used in the process, providing a solvent ratio of ∼1:
1
.5 (MTBE/ether, v/v). The stirring issue typically observed
5
h and further allowed to crystallize at 30 °C for 17 h.
mid-way through this Grignard reaction was significantly
alleviated by carrying out the reaction at higher temperature
-15 to -5 °C) to generate a stirrable slurry that was readily
Gratifyingly, when filtration was carried out at 30 °C, the
unwanted enantiomer was completely purged out and the
desired (-)-6-cis enantiomer was isolated with >99% ee and
(
amenable to the inverse quench with aqueous HCl to
maximize the amount of (()-6-cis. This quench was con-
veniently carried out by adding the reaction mixture slurry
to a cold (0-5 °C) 2.0 M aqueous HCl solution.
To isolate (()-6-cis, the quenched reaction mixture layers
were first separated, the pH of the aqueous layer was adjusted
to pH 10 with ammonium hydroxide, and the crude free base
was extracted into dichloromethane. Subsequently, a solvent
3
4% overall mass recovery after conversion of the salt to
the free base.
These conditions were readily scaled up and reproducibly
implemented on a 214 g scale with a minor modification.
Since (()-6-cis was less soluble in methanol at ambient
(
11) Based on the amount of (()-6-cis isomer present in the quenched reaction
mixture, approximately 70-75% of the theoretical amount was isolated in
a single crop.
Vol. 10, No. 5, 2006 / Organic Process Research & Development
•
917

temperature than (-)-dibenzoyl-L-tartaric acid, it was deemed
operationally more convenient to prepare a slurry of (()-6-
cis and heat it to 65 °C to obtain a clear solution and then
dropwise add a solution of the resolving agent in methanol.
Subsequent slow cooling to 30 °C and further crystallization
at this temperature over 17 h afforded 36% mass recovery
with 99% ee.
Scheme 2
The dibenzoyl-L-tartrate salt was converted to the corre-
sponding free base by treatment with saturated aqueous
sodium bicarbonate. Extraction of the free base with di-
chloromethane followed by solvent exchange into heptanes
and crystallization afforded a slurry of elongated needles that
could be readily isolated by filtration, with >99% ee and
31% overall mass recovery from (()-6-cis.
Thermodynamic Epimerization of (-)-6-cis and Isola-
tion of (+)-CTDP 31,446 as the Hydrochloride Salt. The
thermodynamic epimerization of the resolved (-)-6-cis with
sodium methoxide in refluxing methanol worked well and
did not require any significant optimization. This reaction
was typically conducted in 7 volumes of methanol using 1
molar equiv. of sodium methoxide and proceeded to >98%
conversion within 6 h, as determined by GC analysis. In the
original procedure, the free base derived from the thermo-
dynamic epimerization reaction was isolated as a viscous
oil by stripping the batch to dryness and redissolving the
residue into dichloromethane, followed by washing the
resulting solution with water and further stripping the extracts
to dryness. The final HCl salt formation and crystallization
was then carried out by precipitating the salt from methanol
upon treatment of the crude free base solution with anhydrous
HCl, typically affording moderate yields of CTDP-31,446.
To improve the mass recovery, diethyl ether was added as
an antisolvent on small scale runs, and the product was
isolated in multiple crops.
and easily filtered solid (75-80% yield in a single crop,
Scheme 2).
Conclusions
The original synthetic approach to enantiomerically pure
CTDP-31,446 was optimized and streamlined to define
processing parameters that are more amenable to pilot scale
production. First, the use of diethyl ether was minimized and
restricted to only the amount present in the commercial
Grignard solution. At the same time, MTBE was successfully
incorporated as a cosolvent, and a temperature threshold that
allowed the generation of a mobile slurry was defined. This
made possible the implementation of a more concentrated
process and an inverse quench strategy to maximize the
recovery of the crystalline and easily resolved (()-6-cis
isomer. The resolution procedure was optimized to com-
pletely remove the unwanted enantiomer in a single pass.
Finally, isopropyl acetate was identified as a suitable solvent
not only for the extraction of the resolved (+)-6-trans but
also as an effective antisolvent in a binary crystallization
solvent system to isolate CTDP-31,446 with good mass
recovery and high purity.
Due to efficiency and safety concerns, an alternative
procedure for the final hydrochloride salt formation and
crystallization more amenable to pilot scale production that
avoided or minimized the use of diethyl ether was desirable.
Advantage was taken of the observation that isopropyl acetate
was an excellent solvent for extraction of the free base (()-
6-trans but a rather poor solvent for the HCl salt. These
characteristics could therefore be used advantageously to
further streamline the process by telescoping the final salt
formation in the solvent used for the free base extraction.
Indeed, initial pilot experiments to crystallize the hydro-
chloride salt directly from crude isopropyl acetate extracts
afforded the salt in high yields in a single crop and with
excellent purity. Minimal optimization was required, other
than a controlled addition of the HCl solution to the free
base at 0 °C to minimize the exotherm due to the salt
formation and crystallization. In the optimized process, the
free base (()-6-trans derived from the thermodynamic
epimerization was first extracted into isopropyl acetate after
distillation of methanol, washed with water, clarified by
polish filtration, and cooled to 0-5 °C. A solution of
anhydrous HCl in methanol (1.5 equiv of HCl) was then
slowly added to the free base while maintaining the batch
temperature below 5 °C to afford CTDP-31,446 as a colorless
Experimental Section
General. Reagents purchased from commercial sources
were used as received. Solvents for reactions and isolations
were reagent-grade and used without purification. Proton and
carbon nuclear magnetic resonance spectra were obtained
on a Bruker AVANCE 300 spectrometer at 300 MHz for
918
•
Vol. 10, No. 5, 2006 / Organic Process Research & Development

proton and at 75 MHz for carbon. Tetramethylsilane was
used as an internal reference for the proton spectra, and the
solvent peak was used as the reference peak for the carbon
spectra. Mass spectra were obtained on a Finnigan LCQ Duo
LC-MS (APCI) mass spectrometer. IR spectra (ATR) were
obtained on an Avatar 370 FT-IR Thermo Nicolet. DSC
analysis was obtained on a Mettler Toledo DSC822. HPLC
analysis for chemical purity was performed using a YMC
ODS-AQ (4.6 mm × 250 mm, 5 µm, 120 Å) C18 column,
using a gradient elution with a mobile phase A (0.05%
methanesulfonic acid in water) and mobile phase B (0.04%
methanesulfonic acid in methanol) from 98:2 to 10:90 (A:
B, v/v) and detection at 221 nm. HPLC analysis for chiral
purity was carried out using a Chiralpak AD-H (4.6 mm ×
70 °C. The batch was then allowed to naturally cool to
ambient temperature and further aged for 16 h. The resulting
solids were collected by filtration and dried in a vacuum
oven (40 °C) to afford 322 g of product (53% isolated yield;
1
cis/trans ratio, 12:1) as a tan solid. H NMR (()-6-cis
(CDCl , 300 MHz) δ 1.80 (dq, J ) 12.5, 3.1 Hz, 1H), 2.07
3
(dt, J ) 11.2, 2.9 Hz, 1H), 2.28 (s, 3H), 2.36 (dd, J ) 11.6,
3.7 Hz, 1H), 2.66 (dq, J ) 11.8, 3.7 Hz, 1H), 2.79 (dt, J )
11.8, 3.9 Hz, 1H), 2.93-3.03 (m, 2H), 3.19 (dd, J ) 11.5,
1.5 Hz, 1H), 7.16-7.31 (m, 4H). The filtrate was concen-
trated to give 231 g of a brown oil which contained a 1:6
mixture of (()-6-cis/(()-6-trans.
Kinetic Epimerization of (()-Methyl 4â-(4-Chloro-
phenyl)-1-methylpiperidine-3â-carboxylate, (()-6-trans.
A 5-L reactor equipped with an overhead mechanical stirrer
and a thermocouple was charged under a nitrogen atmosphere
with the filtrate residue containing a mixture of (()-6-cis
and (()-6-trans (1:6 ratio by GC analysis, 231 g, 817 mmol)
and anhydrous THF (2 L). The resulting solution was cooled
to -70 °C, and LDA (477 mL, 2.0 M in THF, 954 mmol,
1.2 equiv) was charged dropwise over 1 h. The reaction
mixture was stirred for an additional hour, and then the
reaction was quenched by transferring the mixture into a
quench vessel containing ice-cold 2 M aqueous HCl (1 L)
with vigorous stirring. The resulting mixture was transferred
back into the reactor, and THF was removed by distillation
under reduced pressure. The aqueous residue was washed
with methyl tert-butyl ether (1 × 1 L), chilled to 0 °C, made
alkaline (approximately pH 10) with concentrated ammonium
hydroxide (200 mL), and extracted with dichloromethane (3
× 1 L). The combined dichloromethane extracts were
distilled under reduced pressure to ca. 500 mL, diluted with
heptanes (500 mL), and further distilled to remove the bulk
of the residual dichloromethane. The final solution in
heptanes was allowed to naturally cool to ambient temper-
ature and age for 18 h. The crystallized solids that formed
were collected by filtration and dried under a vacuum (40
°C) for 24 h to afford 128.6 g of product (56% recovery;
cis/trans ratio, 8:1) as a tan solid.
Resolution of (()-Methyl 4â-(4-Chlorophenyl)-1-
methylpiperidine-3â-carboxylate, (()-6-cis. A 5-L reactor
equipped with an overhead mechanical stirrer, a reflux
condenser, and a thermocouple was charged with (()-methyl
4â-(4-chlorophenyl)-1-methylpiperidine-3â-carboxylate, (()-
6-cis (214 g, 0.80 mol), and methanol (1.8 L, 8.4 volumes).
The resulting slurry was heated to ca. 65 °C, and a solution
of (-)-dibenzoyl-L-tartaric acid monohydrate (301 g, 0.8 mol,
1.0 equiv) in methanol (1.5 L, 7.0 volumes) was added at
such a rate as to maintain the batch temperature above 60
°C (required 30 min). The resulting slightly turbid solution
was then cooled to 30 °C over 5 h at a rate of 7 °C/h and
aged at this temperature for 17 h. The crystallized salt was
harvested by filtration at 30 °C, and the wet cake was washed
with methanol (214 mL, 1 volume), conditioned at ambient
temperature for 30 min, and further dried under a vacuum
(25 in. Hg) at 40 °C for 14 h to afford the dibenzoyl tartrate
salt as a white solid (181.5 g, 36.3% yield). The salt (181 g)
was taken up in 5% aqueous bicarbonate (1.2 L, pH adjusted
250 mm, 5 µm) column, using an isocratic elution with
n-heptane/ethanol (98.5:1.5, v/v) modified with 0.05% di-
ethylamine, and detection at 221 nm. GC analysis was
conducted using a Restek RTX-5 (30 m × 0.32 mm × 0.25
µm) capillary column fitted to a Hewlett-Packard 5890 gas
chromatograph interfaced with a Hewlett-Packard 3396A
integrator. The analysis conditions were injector 250 °C,
carrier gas helium, splitless injection (1 µL), oven temper-
atures 80 °C, isothermal for 4 min, ramp at 15 °C/min to
250 °C, hold at 250 °C for 10 min, and detector FID at 250
°C. Retention times: (()-6-trans 20.6 min and (()-6-cis 21.7
min. Elemental analysis was performed by Quantitative
Technologies, Inc.
Preparation of (()-Methyl 4â-(4-Chlorophenyl)-1-
methylpiperidine-3â-carboxylate, (()-6-cis. A 12-L reactor
equipped with an overhead mechanical stirrer and a thermo-
couple was charged with a 1.0 M solution of p-chloro-
phenylmagnesium bromide in diethyl ether (3.1 L, 3.1 mol,
1.4 equiv) under a nitrogen atmosphere. The Grignard
solution was cooled to -20 °C, and then a solution of
arecoline (350 g, 2.26 mol) in anhydrous methyl tert-butyl
ether (1.8 L, 5.1 volumes) was charged dropwise over 1 h
while maintaining the batch temperature below -5 °C. The
batch was vigorously stirred throughout the addition of the
arecoline solution, and the reaction progress was monitored
by GC analysis. After further reaction for 2 h, GC analysis
indicated <5% residual arecoline. The reaction was quenched
by carefully transferring the slurry into a quench vessel
containing ice-cold 2 M aqueous HCl (3.5 L), with vigorous
stirring. A small amount of residual reaction mixture in the
reactor was quenched with an additional 1 L of ice-cold 2
M aqueous HCl, and the wash was combined with the bulk
of the quenched reaction mixture. The resulting biphasic
mixture was allowed to stir for 15 min, the layers were
separated, and the organic layer was washed with ice-cold 2
M aqueous HCl (1 × 1 L). The combined aqueous layers
were chilled to 0 °C, carefully made alkaline (approximately
pH 10) with concentrated aqueous ammonium hydroxide (ca.
6
2
50 mL) and extracted with dichloromethane (1 × 2 L, then
× 1 L). The combined dichloromethane extracts were
concentrated by distillation under reduced pressure to ap-
proximately 1 L of residual solution. Heptanes (1 L) were
added to the residual methylene chloride solution, and
distillation was continued, with a final batch temperature of
Vol. 10, No. 5, 2006 / Organic Process Research & Development
•
919

to pH 10 with NaOH) and dichloromethane (0.9 L), and the
layers were separated. The aqueous layer was further
extracted with dichloromethane (1 × 0.9 L), and the
combined organic layers were washed with water (1 × 0.9
L) and then distilled to a residual volume of approximately
a thermocouple. The crude free base solution was cooled to
0 °C, and then a solution of anhydrous HCl in methanol (ca.
9 wt %, 7 mg/mL, 175 mL) was added dropwise over 30
min while maintaining the batch temperature below 5 °C.
The resulting slurry was allowed to age at 0 °C for 1 h and
then filtered. The wet cake was washed with 20% (v/v)
methanol in isopropyl acetate (1 × 60 mL), conditioned
under a vacuum at ambient temperature for 2 h, and further
dried under a vacuum (25 in. Hg) at 40 °C for 45 h to afford
CTDP 31,446 as a white crystalline solid [55.4 g, 81.4%
0.5 L. To the residue were added heptanes (750 mL), and
the resulting mixture was further distilled under reduced
pressure to a residual volume of approximately 300 mL. The
resulting slurry was aged at ambient temperature for 1 h,
filtered, and dried under a vacuum (25 in. Hg) at 40 °C for
25
12 h to afford (-)-methyl 4â-(4-chlorophenyl)-1-methyl-
yield from (-)-6-cis]. Mp: 246.5 °C. [R]
D
+56.2 ° (c1.03,
piperidine-3â-carboxylate, (-)-6-cis [67 g, 31.3% yield from
EtOH). HPLC assay for chemical purity: 99.8% (AUC).
HPLC assay for chiral purity: 99.9% ee. H NMR (DMSO-
1
(()-6] as colorless needles.
Thermodynamic Epimerization of (-)-Methyl 4â-(4-
Chlorophenyl)-1-methylpiperidine-3â-carboxylate, (-)-
-cis, and Preparation of CTDP 31,446. A 1-L reactor
equipped with an overhead mechanical stirrer, a reflux
condenser, and a thermocouple was charged with a slurry
of (-)-methyl 4â-(4-chlorophenyl)-1-methylpiperidine-3â-
carboxylate (60 g, 0.224 mol) in methanol (420 mL, 7.0
volumes), followed by sodium methoxide (12.1 g, 0.224 mol,
6
d , 300 MHz) δ 1.92 (d, J ) 12.6, 1H), 2.22 (dq, J ) 12.9,
2.9 Hz, 1H), 2.78 (s, 3H), 2.99 (dt, J ) 12.0, 3.7 Hz, 1H),
3.06-3.26 (m, 2H), 3.31-3.52 (m, 2H), 3.36 (s, 3H), 3.64
(d, J ) 11.1 Hz, 1H), 7.22 (d, J ) 8.2 Hz, 2H), 7.43 (d, J
6
1
3
) 8.4 Hz, 2H), 11.56 (br s, 1H). C NMR (DMSO-d
41.7, 42.6, 45.7, 52.2, 53.1, 53.8, 129.0, 129.3, 132.0, 141.1,
171. Anal. Calcd for C14 ‚HCl: C, 55.27; H, 6.29;
6
) 29.7,
H
18ClNO
2
N, 4.60; Cl, 23.30. Found: C, 55.29; H, 6.34; N, 4.50; Cl,
+
1
.0 equiv). The resulting mixture was then heated to 60 °C,
23.10. MS m/z (APCI): 268 [M + H] .
and the reaction was allowed to proceed for 6 h, whereupon
GC analysis of an aliquot indicated <2% residual (-)-6-cis
isomer. The bulk of methanol was removed by distillation
Acknowledgment
We are grateful to Dr. C. Jamie Biswas (National Institute
on Drug Abuse, Bestheda, MD) and Dr. Miles P. Smith
(Molecular Insight Pharmaceuticals, Cambridge, MA) for
helpful discussions. We thank Ping He and Dr. Thomas P.
Gregory (Albany Molecular Research, Inc., Albany, NY) for
analytical methods development and Dr. Robert J. Duguid
and Dr. Karl Reineke (Albany Molecular Research, Inc.,
Albany, NY) for a critical review of the manuscript.
(ca. 410 mL removed), and the residue was treated with a
mixture of isopropyl acetate (450 mL, 7.5 volumes) and
saturated aqueous ammonium chloride (450 mL, 7.5 vol-
umes). The layers were separated, and the aqueous layer was
further extracted with isopropyl acetate (1 × 450 mL). The
combined organic extracts were washed with water (1 × 450
mL, then 1 × 300 mL) and concentrated under reduced
pressure to approximately 360 mL of residual solution. This
solution was polish-filtered through a 1.0 µm filter and
charged into a 1-L reactor equipped with an overhead
mechanical stirrer, a pressure-equalizing addition funnel, and
Received for review June 9, 2006.
OP060114G
920
•
Vol. 10, No. 5, 2006 / Organic Process Research & Development