A scaleable route to the pure enantiomers of verapamil

Article abstract of DOI:10.1021/op000059q

A versatile route to single enantiomer verapamil from readily available raw materials is described. The key intermediate, 4-cyano-4-(3,4-dimethoxyphenyl)-5-methyl hexanoic acid (verapamilic acid), was resolved efficiently with α-methyl benzylamine. Stereochemical integrity at the quaternary carbon centre was preserved through subsequent steps to give either (R)-or (S)-verapamil in good overall yield. This sequence incorporated a selective borane-mediated reduction of a tertiary amide. Process scale-up to the pilot plant has been demonstrated successfully for the resolution step.

Full text of DOI:10.1021/op000059q

Organic Process Research & Development 2000, 4, 467−472
A Scaleable Route to the Pure Enantiomers of Verapamil
Robin M. Bannister, Michael H. Brookes,* Graham R. Evans, Ruth B. Katz, and Nicholas D. Tyrrell
Celltech-Chiroscience Ltd., Granta Park, Great Abington, Cambridge, CB1 6GS, U.K.
Abstract:
A versatile route to single enantiomer verapamil from readily
available raw materials is described. The key intermediate,
4-cyano-4-(3,4-dimethoxyphenyl)-5-methyl hexanoic acid (vera-
pamilic acid), was resolved efficiently with r-methyl benzy-
lamine. Stereochemical integrity at the quarternary carbon
centre was preserved through subsequent steps to give either
(R)-or (S)-verapamil in good overall yield. This sequence
incorporated a selective borane-mediated reduction of a tertiary
amide. Process scale-up to the pilot plant has been demonstrated
successfully for the resolution step.
acceptable cost. The substitution pattern at the chiral
quaternary centre cannot be readily derived from members
of the chiral pool or by asymmetric induction. Efforts to do
so have resulted in lengthy syntheses comprising many
functional group interconversions, as well as separations of
diastereomeric intermediates by column chromatography.6-8
Verapamil free base has been resolved directly, but to obtain
high enantiomeric excesses (ee’s) required inefficient, mul-
tiple recrystallisatons or very expensive auxiliaries.9,10 Fur-
thermore, no salt of verapamil has been identified as forming
a conglomerate, and bulk synthesis of the parent amine
involves only achiral precursors.11 McCague et al devised
an adaptation of the latter that was based around resolution
of a secondary amine intermediate, but this tactic distanced
the overall route from readily available starting materials.12
Another approach was based upon the resolution of 4-cyano-
4-(3,4-dimethoxyphenyl)-5-methyl hexanoic acid (“verapa-
milic acid”) (2), as the key intermediate.13 The route was
unattractive, as described, because it employed the expensive
and toxic alkaloid brucine for the resolution step. Addition-
ally, downstream conversion to verapamil involved a low-
yielding reduction step and a high-temperature coupling
reaction, with neat reactants, whilst purification of the final
product was both laborious and indirect. However, this
approach shared key feedstocks with manufacture of the
racemic verapamil, allowed flexibility in onward conversion
to the final product, and did not require column chromatog-
raphy for any step. We reasoned that should a better
resolution be developed for 2 then a means to meet our
objectives would follow. Herein, we describe our efforts to
Introduction
Verapamil (1) has a well-established place in the treat-
ment of cardiovascular ailments.1 The compound is usually
administered as its hydrochloride salt. Currently there is
renewed interest in enhancing this drug’s portfolio by
replacing the racemate with either the pure single enanti-
omers, or non-stoichiometric mixtures thereof. These isomers
differ significantly in their biological effects and pharma-
codynamics.2 (S)-(-)-Verapamil has quite specific trans-
membrane calcium channel antagonist activity, whilst its
antipode influences a wider range of cell pump actions,
including that for sodium ions. Formulations enriched in the
(S)-enantiomer are likely to confer greater pharmacological
efficacy, with diminished side effects, for cardiac arrhythmia
and hypertension. In contrast, angina appears to be better
alleviated by (R-(+)-verapamil. Unrelated therapeutic targets,
such as multidrug resistance in cancer chemotherapy,3 and
migraine,4 have also been proposed for verapamil. These
more speculative outlets may also benefit from an optimi-
sation of the enantiomer ratio.5
Lack of a scalable “chiral switch” for verapamil has
hindered investigations since, hitherto, there has been insuf-
ficient material available to support detailed clinical trials.
A number of syntheses have been published, but these have
little prospect of providing kilograms of bulk active at an
* Author for correspodence. E-mail: michaelbrookes@chiroscience.com.
(1) Cervoni, P.; Crandall, D. L.; Chan, P. S. Cardiovascular Agents. In Kirk-
Othmer Encyclopedia of Chemical Technology, 4th ed.; Kroschwitz, J. L.,
Executive Ed.; Howe-Grant, M., Ed.; Wiley: New York, 1992; Vol. 5, pp
207-300 and references therein.
(2) Longstreth, J. A. Verapamil: A Chiral Challenge to the Pharmacokinetic
and Pharmacodynamic Assessment of Bioavailability and Bioequivalence.
In Clinical Pharmacology, 2nd ed.; Wainer, I. W., Drayer, D. E., Eds.;
Drug Stereochemistry, Vol. 11; Dekker: New York, 1993; p 315.
(3) Eliason, J. F.; Ramuz H.; and Kaufmann, F. Int. J. Cancer. 1990, 46, 113.
(4) Peroutka, S. J. Ann. Neurol. 1988, 23, 500.
(6) Ramuz, H. HelV. Chim. Acta. 1975, 58, 2050.
(7) Theodore, L. J.; Nelson, W. L. J. Org. Chem. 1987, 52, 1309.
(8) Gilmore, J.; Prowse, W.; Steggles, D.; Urquhart, M.; Olkowski J. J. Chem.
Soc., Perkin Trans. I 1996, 2845.
(9) Blasche, G.; Doebber, P. Ger. Patent DE 3723684.
(10) Ehrmann, O.; Nagel, H.; Karau, W. Int. Patent Appl. WO9316035 and
WO9408950.
(11) Dengel, F. Belg. Patent 615,861, 1962 ; Idem. U.S. Patent, 3,261,859, 1966.
(12) McCague, R.; Wang, S. Int. Patent Appl. WO 9509150.
(13) Trieber, E. J.; Raschack M.; and Dengel, F. Ger. Patent 2,059,923, 1972 .
(5) Longstreth, J. A.; Fakhreddin, J. Int. Patent Appl. WO9744025A1 (U.S.
Patent 5,955,500, 1999).
10.1021/op000059q CCC: $19.00 © 2000 American Chemical Society and The Royal Society of Chemistry
Published on Web 08/19/2000
Vol. 4, No. 6, 2000 / Organic Process Research & Development
•
467

develop a process with the potential to deliver large quantities
of pure, single isomer verapamil.
the latter was recovered from the aqueous phase. Verapamilic
acid was isolated in 89-96% physical yield and 97-99%
HPLC purity after controlled precipitation by acidification
at 60-65 °C.
Results and Discussion
The first hurdle to overcome was to secure a reliable
procedure to make racemic 2. This compound is accessible
by Michael addition of either acrylonitrile or acrylate esters
to 2-isopropyl homoveratronitrile (3), as shown in Scheme
. Since 3 is already an intermediate in the commercial
synthesis of racemic verapamil, it is available in bulk at low
cost. Knoll Pharmaceuticals supplied our 2-isopropyl ho-
moveratronirile, which was used without purification. Reac-
tion of 3 with acrylonitrile proceeded best in tert-butyl
alcohol with a catalytic amount of potassium tert-butoxide
as base. After some experimentation, a loading of 0.5 mol
% of this base was adopted. HPLC reaction monitoring,
which allowed the reagent charges to be adjusted prior to
the water quench, assured essentially quantitative conversion
to bis-nitrile (4). Although 4 could be isolated in high yield
The alternative pathway via methyl acrylate proved
unsatisfactory. Under the best conditions tried, a 2:1 mixture
of esters 7 and 8 was obtained, which was contaminated by
solvent (tert-butyl alcohol) addition products for example
tBuOCH2CH2CO2Me. Attempts to perform the reaction in
THF/tert-butyl alcohol mixtures led to the bis-addition
product(s) 9 appearing in significant proportions.
A small screen of eight resolving agents was undertaken
for the resolution of 2, of which three were effective. The
particulars of the lead results are highlighted in Table 1. (R)-
Quinine, the least efficient of the base trio, also shares many
of the disadvantages of brucine (expense, availability of only
the “natural” enantiomer, and relatively high molecular
weight). These two alkaloids give complementary resolutions
to the enantiomers of 2. In contrast, 1-naphthylethylamine
and R-methyl benzylamine (R-MBA) are available in either
enantiomeric form, which allows access to both 2a and 2b
with equal ease. However, the cost of chiral 1-naphthylethy-
lamine precludes it from further consideration for large-scale
work. Hence, cheap and commercially readily available
single isomer R-MBA was identified as the best resolving
agent for 2, in terms of both cost and efficiency (highest S
factor).14
Scheme 1
Further development work confirmed that the highest
diastereomeric excesses were obtained when the R-MBA
salts of 2 were crystallised from ethyl acetate. Processing
constraints with this solvent were imposed by formation of
N-acetyl R-MBA, which became significant above 60 °C.
Toluene was an inefficient solvent for the resolution (S )
0.55), but proved ideal for purification of enantiomerically
enriched streams of 2, generated from it. This solvent
conferred greater process robustness, being more chemically
inert and allowing any moisture carry-over to be easily
removed azeotropically. Entrained water had a demonstrable
adverse affect on the crystallisation. Careful temperature
control during cooling back to 0-5 °C, prior to isolation,
was also very important in securing the best crystal forms
and diastereomeric excesses (de’s) from both ethyl acetate
and toluene.
Scheme 2 depicts the overall resolution process we
devised, whereby both enantiomers of 2 were obtained
sequentially in >95% de from single crystallisations, each
in ∼40% physical yield. If desired, salts 10a and 10b could
be purified to 98-99% de by recrystallisation from ethyl
acetate. The latter de’s were obtained by a single crystalli-
sation in the laboratory on some occasions, and it may be
that more precise control of the vessel jacket temperature
during the early stages of cooling back will allow this feature
to be scaled up too. On our pilot plant the process was run
as shown in Scheme 2, but there is no reason 10b could not
be crystallised first, by transposing the addition order of the
R-MBA enantiomers. Reuse of the verapamilic acid remain-
after crystallisation from methanol (mp 97.2-100.1 °C), we
pursued a procedure that incorporated the subsequent (pri-
mary) nitrile hydrolysis. After drowning out in water, the
tert-butyl alcohol was distilled out and replaced with ethanol
and aqueous potassium hydroxide. The mixture was reheated
to distill out the ethanol and then boiled under reflux until
the conversion to verapamilic acid was >90% by HPLC
(16-18 h). A homogeneous medium was necessary to start
conversion of 4 to 5 at a satisfactory rate. Ethanol, but not
tert-butyl alcohol, solubilised 4 to achieve this: hence the
need for two distillation steps. Provision of a phase transfer
catalyst in the future may permit ethanol to be dispensed
with. Although 4 had disappeared after 2-3 h, the persistence
of amide 5 and adduct 6 necessitated long periods at reflux.
Both intermediates were isolated from the hydrolysis and
identified: the former was extracted into ethyl acetate, whilst
(14) Fogassy, E.; Lopata, A.; Faigl, F.; Darvas, F.; Acs, M.; Toke, L.; Tetrahedron
Lett. 1980, 21, 647.
468
•
Vol. 4, No. 6, 2000 / Organic Process Research & Development

Table 1.
resolving agent
solvent
acetone
yield of salt (d.e. %)
S factora
0.46
(R)-(-)-quinine (MW 324)
72%
(84.5%) (2a)
86%
(82.5%) (2b)
88%
(96.6%) (2a)
(R)-(+)-1-(1-naphthyl)- ethylamine (MW 171)
(S)-(-)-R-methyl benzylamine (MW 121)
ethyl acetate
ethyl acetate
0.71
0.76
a S ) efficiency ) diastereomeric excess (de)% of solid × yield × 2.
Scheme 2
Scheme 3
which was redistilled prior to use (bp 109-110 °C at 1
mBar). Direct coupling of resolved 2 with amine 11 (1.1
mol equiv) was achieved in 85-90% yield, in toluene or
xylene solvent, and facilitated by 0.5-1 mol % 3-nitrophenyl
boronic acid (3-NPBA).17 The high efficiency of the latter
reagent at least partially offset its high cost. Boric acid itself
was also effective as catalyst, but required much higher
loadings (10-20 mol %) and gave somewhat lower conver-
sions to verapamilamide (12). After removing excess amine
11 with an aqueous acid wash, the organic solution was
azeotrope-dried and concentrated to crystallise amides 12a
or 12b. Once again, the single isomer forms showed greater
reluctance to crystallise than racemic 12, and seeding with
authentic material was essential for consistency. The amides
crystallised most easily from concentrated xylene solutions
( ∼2.1 volumes). Once the product started to come out of
solution, dry xylene was added to the flask to maintain
suspension mobility.
ing in the toluene mother liquors appears to be feasible, and
would raise the total utilisation of racemic 2 to above 80%.
Whilst racemic 2 was relatively easy to obtain as a solid,
the single isomer forms preferred to oil out under the same
conditions. Because of these difficulties, the decomposition
of 10a or 10b, with aqueous acid, was integrated into
downstream processing, and no further attempt made to
isolate the liberated acids.
Although resolved 2 has been elaborated to verapamil
after reduction to the corresponding alcohol13 or aldehyde,15
no synthetic route via coupling with N-methyl homovera-
trylamine (11) at the same oxidation level has been re-
ported.16 We decided to explore this option, shown in Scheme
3, and assess its feasibility for scale-up. Amine 11 is the
other feedstock shared with the extant commercial synthesis
of racemic verapamil. Uetikon provided our supply of 11,
Interestingly, the proton NMR spectrum of 12 is com-
plicated by the presence of a 1:1 mixture of rotamers. At
ambient temperature, several peaks were doubled up, includ-
(15) Kastner, G.; Hardo, S.; Geiss, K. H. U.S. Patent 4,350,636, 1982.
(16) Bannister, R. M.; Evans, G. R.; Skead, B. M. Int. Patent Appl. WO 9729081.
(17) Ishihara, K.; Ohara, S.; Yamamoto, H. J. Org. Chem. 1996, 61, 4196.
Vol. 4, No. 6, 2000 / Organic Process Research & Development
•
469

ing the N-methyl group, which gave two singlets at (δ 2.73
and 2.80). These resonances coalesced at ∼80 °C, indicating
that a significant barrier to free rotation about the amide bond
exists. The HPLC chiral purity of 12 could not be checked
accurately, since only partial separation of the enantiomer
peaks was obtained, even on a bovine serum albumen (BSA)
protein or Chirobiotic T columns. This deficiency was
tolerable because fully effective methods were in place for
both 10 and verapamil. Further process refinement may allow
uncrystallised amide to be charged directly to the reduction
step.
The final reduction of 12 was the most chemically
challenging step of our route. A wide variety of reagents
are capable of converting amides to amines,18 but our
requirement for selectivity imposed significant constraints
on the choice. Scouting experiments discounted use of
aluminium hydride reagents because they tended to be too
aggressive, causing some cleavage of the amide itself. In
our hands, the action of borane:THF was confined to
reduction, but it failed to discriminate between the amide
and nitrile groups to a satisfactory degree; hence, a significant
proportion of 12 ended up as the diamine 13.7 Borane:
dimethyl sulphide complex (BMS) was able to reduce the
amide functionality completely, without affecting the nitrile
group substantially, and was advantageous in terms of its
far greater stability.19 Use of BMS also allowed a choice of
reaction solvent. Robust conditions for reducing 12 in 5 vols
of THF with 1.85 equiv of BMS were quickly identified.
Although reduction of an amide consumes only two hydride
equivalents formally, a second mole of borane complexes
to the amine formed; 1.85 equiv therefore represents a 10%
mole excess of theory. Over-reduction to 13 was still very
limited (10.8 mol %) even after stirring for 60 h with 2.5
equiv of this reagent. 1,2-Dimethoxyethane (DME) provided
an equally suitable reaction medium, but 8 vols were required
to dissolve 12. HPLC monitoring showed that the reaction
went to completion, and an in-process assay was consistent
with 85-90% active ingredient yield of verapamil.
Whilst conditions for the reduction itself were facile, the
work-up and product purification required detailed optimi-
sation. Forcing conditions were needed to decompose the
verapamil:borane complex and aqueous or methanolic solu-
tions of hydrogen chloride at reflux were the most effective.
Both methanolic and the more concentrated hydrochloric acid
solutions (3-6 M) appeared to promote side reactions which
significantly depressed the quality and physical yield of the
product. We were able to suppress, but not eliminate, this
deterioration by using 1 M hydrochloric acid in the step.
The extent of these side reactions was hard to quantify since
their high molecular weights or polarities caused them to be
retained on the GC and HPLC columns used for reaction
monitoring. In our experience, boiling with lower-strength
acid for longer periods gave the best tradeoff, in terms of
crude product yields. Despite these measures, the level of
impurities was such that most of the product oiled out upon
cooling the aqueous solution. This phase was readily miscible
with dichloromethane but would not dissolve in ethyl or
isopropyl acetate.
Most literature verapamil preparations have isolated the
free base, purified it, then appended conversion to, and then
crystallisation of, the hydrochloride salt. This approach is
disadvantageous for scale-up because of the variety of
solvents employed and the lengthy processing it entails. Our
shorter work-up exploited the solubility of verapamil hy-
drochloride in dichloromethane, which also allowed it to be
extracted directly from the acid quench liquors. Apart from
giving better volume efficiency, more polar impurities, such
as 13, which would also have been extracted following
basification, remained in the aqueous phase.
Attempts to crystallise the residue obtained from evapo-
rating the dichloromethane extracts, by existing methods,
were largely unsuccessful due to the impurity burden. We
found that a manageable solid precipitated if a concentrated
solution of the residue in 2-propanol was added slowly to a
large volume of well-stirred methyl tert-butyl ether, which
had been seeded with authentic 1a/b. Although a 73% (net)
physical yield was obtained, this material still required a
recrystallisation from 10 vols of 2-propanol to raise its
specification to >99.5% PAR (HPLC) and mp >130 °C.20
The latter step proceeded with 85% recovery, making the
Stage’s overall yield 62.1%. Sticky, inferior crystals pre-
cipitated after an attempt to substitute a concentrated
dichloromethane solution for one of 2-propanol in the first
crystallisation. Optimisation work continues on this step, with
a view to obviating the need for a recrystallisation.
Despite the convenience of the reduction procedure, it is
still only suitable for large laboratory or small kilo laboratory
scale, due to the use of BMS. The disadvantages of this
reagent are its expense and the need for specialised handling
equipment, when working at higher volumes. We have
therefore tried to substitute borane generated in situ (from
NaBH4 and HCl) for this transformation. A change in
reaction solvent from THF to DME was necessary to
accommodate anhydrous hydrogen chloride. Although this
method is presently less clean than BMS mediated reduction,
we are confident that further development work will deliver
a satisfactory process.
Conclusions
We believe that our endeavours have provided the basis
for a simple and efficient process to both (R)- and (S)-
verapamil hydrochloride which could be run at up to kilo
laboratory scale, with little further modification. The overall
yield of single isomer verapamil hydrochloride from 3 stands
at 35-40% of theory for each enantiomer. There are only
four steps to the synthetic route. Reuse of the toluene liquors
from the resolution, improvements to the efficiency of the
amide reduction work-up, and final product crystallisations
will probably improve this figure. Our synthetic strategy
should permit the preparation of related compounds in single
isomer form, such as the pharmacologically active metabolite
(18) Pelter, A.; Smith, K.; Brown, H. C. In Borane Reagents; Katritsky, R.; Meth-
Cohn, O.; Rees, C. W. Eds.; Academic: San Diego, 1988.
(19) Seyden-Penne, J. Reductions by the Alumino- and Borohydrides in Organic
Synthesis; VCH: New York, 1991.
(20) For a thorough review on analytical topics related to verapamil see: Chang,
Z. L. Analytical Profiles of Drug Substances 1988, 17, 643.
470
•
Vol. 4, No. 6, 2000 / Organic Process Research & Development

crystal growth was evident. Once this had occurred, the pH
was reduced to <1.0 with further 6 M hydrochloric acid (10
L), and the resulting suspension cooled to 20 °C, prior to
filtration. The solid was washed with water (3 × 12 L) before
drying under vacuum at 50-60 °C to afford 2 as an off-
white microcrystalline powder (9.7 kg, 33.4 mol, 91.3%
yield,) mp 113.7-116.8 °C. 1H NMR spectrum (CDCl3): δ
0.83 (d, 3H); 1.22 (d, 3H); 2.2 (m, 3H); 2.5 (m, 2H); 3.89
(s, 3H); 3.9 (s, 3H); 6.9 (m, 3H); 13.0 (b,s 1H). HPLC purity
99.0% PAR; method: column NovaPak C18, 4 µm, 3.9 ×
150 mm, mobile phase 40 MeCN, 59 H2O, 1.0 trifluoroacetic
acid (TFA), flow rate 1 mL/min, temperature 40 °C, detector
UV at 230 nm. Retention times: 3, 6.0 min; 4, 5.6 min; 2,
3.1 min; 5, 1.9 min; 6, 1.0 min. IR spectrum (KBr disk):
2230.1(w), 1705.2(s) cm-1.
Resolution of Verapamilic Acid. (R)-Verapamilic Acid
(S)-r-MBA Salt (10a). (S)-R-MBA (3.3 kg, 27.3 mol) was
added over 10 min to a solution of 2 (7.93 kg, 27.3 mol) in
ethyl acetate (35.7 kg, 40.0 L) at 40 °C. The batch
temperature was then adjusted to 50 °C prior to seeding with
salt 10a (0.03 kg). After a 2 h stir-out at 50 °C, the
preparation was cooled slowly to 0-5 °C, at a rate of no
more than 5 °C per hour. Following 1 h at 0-5 °C, the
suspension was filtered and the cake washed with cold ethyl
acetate (2 × 15 kg, 13.5 L). The solid was dried to constant
weight at 35-45 °C in vacuo to give salt 10a as a white
norverapamil (14),2 as well as analogues gallopamil (15),7
emopamil (16),8 and other structural relatives. Since the
demand for chiral verapamil hydrochloride is likely to be
skewed in favour of the (S)-enantiomer, we are presently
examining a means to recycle the unwanted isomer. Disas-
sembly of the quarternary carbon centre to facilitate race-
misation is not trivial, but we have demonstrated the
feasibilty of a retro-Michael reaction on 2.21 It is anticipated
that further process development will allow the amide
reduction step to be run in larger, more general purpose, plant
equipment.
Experimental Section
Proton NMR spectra were obtained from a Bruker AM
400-MHz machine. A VG Platform II spectrometer was used
to record mass spectra. A Perkin-Elmer 1600 FTIR spec-
trometer provided the infrared data. Gas chromatography
analysis was performed using a Perkin-Elmer Autosystem
600 model in conjunction with a DB-5 column. Melting
points were determined with either a Leica Galen (III)
microscope hot stage or a Mettler-Toledo DSC 12E, and are
uncorrected. Optical rotations were measured at 589 nm on
a Perkin-Elmer polarimeter. CHN combustion data was
obtained from a Leeman Labs Inc CE440 elemental analyser.
All laboratory reagents were used as supplied, unless stated
otherwise. The reactions described below were carried out
under a nitrogen atmosphere and routinely monitored for
completion by GC or HPLC. Process water was deionised
throughout. The xylene solvent used was the commercial
mixture of isomers and ethyl benzene.
Verapamilic Acid (2). Potassium tert-butoxide (0.2 kg,
1.78 mol) was charged to a stirred solution of nitrile 3 (8.0
kg, 36.7 mol) in tert-butyl alcohol (31.0 kg, 40 L). The vessel
contents were adjusted to 20-25 °C, and then acrylonitrile
(2.53 kg, 47.7 mol) was introduced over 2.5 h. After this
temperature range was maintained for a further hour, the
reaction was sampled for a completion check and then
quenched with water (1.0 L). A solvent exchange was
performed by distilling out the tert-butyl alcohol at atmo-
spheric pressure whilst feeding in water to keep the batch
volume approximately constant. Ethanol (16.0 L) and a
solution of potassium hydroxide (3.1 kg, 55.4mol) in water
(8.0 L) were then added, and distillation was repeated until
the pot temperature reached 100 °C. The batch was subse-
quently boiled under reflux until conversion to product was
>90% by HPLC (18 h). The solution was cooled to 50-60
°C prior to careful addition of 6 M hydrochloric acid (5.5
L) to obtain the cloud point (at ∼pH 6). The preparation
was seeded with authentic 2 (0.08 kg), and stirred out until
crystalline solid (4.35 kg, 10.6 mol, 38.8%), mp 132.4-134
1
°C, [R]20 ) +15.4° (c 5.2, MeOH). H NMR (D2O): δ
D
0.71 (d, 3H); 1.14 (d, 3H); 1.33 (d, 3H); 1.63 (m, 1H,); 1.98
(m, 1H); 2.18 (m, 1H); 2.29 (m, 2H); 3.8 (s, 6H); 4.17 (q,
1H); 6.93-7.02 (m, 3H); 7.28 (m, 1H); 7.38 (dd, 2H); 7.45
(d, 2H). HPLC: 99.0% PAR; 96.7%ee. Method: column
Chiralcel OJ 10 µm, 4.6 × 250 mm, mobile phase 88
heptane, 11.5 2-propanol, 0.5 TFA, flow rate 1.0 mL/1 min,
temperature 30 °C, detector at 230 nm, retention times
R-MBA, 5 min; N-acetyl-R-MBA, 8.7 min; 2a, 10.6 min;
2b, 15.2 min; dimethoxy isobutyrophenone (feedstock im-
purity) 28 min. Calcd for C24H32N2O4; C, 69.88; H, 7.82; N,
6.79%. Found: C, 69.61; H, 7.85; N, 6.83%.
(S)-Verapamilic Acid (R)-r-MBA Salt (10b). The
combined ethyl acetate mother liquors and washes from the
above preparation of (10a) were stirred with 1 M hydro-
chloric acid (18 L) (pH check <2.0) and then water (2 × 10
L). The organic phase was subsequently concentrated by
vacuum distillation (200-250 mBar), until the pot temper-
ature reached 60 °C. After a full solvent swap to toluene,
with concomitant azeotropic removal of entrained water, the
residual solvent charge was adjusted to 4 vols, with respect
to the (calculated) free 2 content. After the solution cooled,
to 50-55 °C, (R)-R-MBA (1.96 kg, 16.2 mol) was added.
The solution was heated back to 60-65 °C prior to seeding
with authentic salt 10b (0.03 kg). After stirring out for 2 h,
the vessel contents were carefully cooled to 5-10 °C over
19 h. The crystallised solid was filtered off, washed with
toluene (2 × 14.0 kg, 16.0 L), and dried in vacuo, as above,
to furnish salt 10b, (4.65 kg, 11.3 mol, 41.4%), (mp 132.2-
1
(21) McCague, R. International Patent Appl. WO 9729080.
133.6 °C, [R]20 ) -14.9° (c 5.0, methanol). H NMR
D
Vol. 4, No. 6, 2000 / Organic Process Research & Development
•
471

spectrum (CDCl3) was identical to that of 10a. HPLC: 99.2%
PAR; ee 96.3%, method as above. CHN Found: C, 69.88;
H, 7.88; N, 6.55%.
to warm to 20-25 °C, whilst stirring overnight (16 h). The
completed reaction was added over 1 h to 1 M hydrochloric
acid (1000 mL), which had been preheated to 80-85 °C.
Provision was made for simultaneous THF removal by
distillation. Once quenching was complete, the pot temper-
ature was raised to 100 °C, and the residual mixture boiled
under reflux for 4 h. After cooling back to 20 °C, the
preparation was extracted with dichloromethane (3 × 300
mL). These extracts were combined, back-washed with water
(50 mL), then solvent swapped to 2-propanol (300 mL). The
2-propanol solution was subsequently added over 1h to a
well-stirred suspension of authentic 1b (10 g) in methyl tert-
butyl ether (1500 mL). The resulting mixture was stirred for
2 h, then filtered and the solid washed with methyl tert-
butyl ether (2 × 100 mL). After drying in vacuo at 50-60
°C, the crude product was obtained as a white powder (76.6
g,22 0.156 mol, 73.0% yield); (mp 126-133 °C). This
material was purified by recrystallisation from 2-propanol
(850 mL) to obtain good quality (S)-verapamil hydrochloride
(65.1 g, 85% recovery (0.133 mol, 62.0% overall) mp 132-
(S)-Verapamilamide (12b). Salt 10b, (494 g, 1.2 mol)
was partitioned between xylene (3.0 L) and 1 M hydrochloric
acid (1.5 L) at 40 °C. After the reaction mixture was washed
further with water (2 × 1.0 L), amine 11 (260 g, 1.32 mol)
and a solution of 3-nitrophenyl boronic acid (1.0 g, 6.0 mmol)
in xylene (0.25 L) were added sequentially, and then the
whole was boiled under reflux for 24 h. A Dean-Stark
apparatus removed water carried over from the separations
and that arising from the condensation. After it was
established that the reaction was complete by HPLC, the
preparation was cooled to, and then maintained at, 40 °C
whilst washing with 1 M hydrochloric acid (1.0 L), water
(1.0 L), 5% sodium bicarbonate solution (1.0 L), and finally
more water (1.0 L). The organic phase was concentrated and
dried by distilling out 2.0 L of wet xylene. Upon cooling
back to 50-55 °C, the residual solution was seeded with
authentic 12b (12.5 g). Once crystallisation had initiated, the
resulting suspension was carefully cooled to 0-5 °C over 3
h, whilst diluting with dry xylene (1.5 L) to maintain
mobility. The crude product was filtered off, washed with
xylene (2 × 0.5 L), and then dried in vacuo at 50 °C to
afford 12b as a white microcrystalline solid (479.8 g, 1.03
134 °C, [R]25 ) -9.0° (c 5.04, ethanol), [lit.7 131-133
D
1
°C, [R]24 ) -8.9°(c 5.03, ethanol)]. H NMR (D2O): δ
D
0.76 (d, 3H), 1.19 (d, 3H), 1.34 (m, 1H); 1.71 (m, 1H); 2.01
(septet, 1H); 2.21 (m, 2H); 2.85 (s,3H); 2.87 (m, 2H); 3.08
(m,1H); 3.20 (m, 1H); 3.26(t, 2H); 3.80-3.87 (4s, 12H); 6.73
(dd, 1H); 6.87 (d, 1H); 6.92-7.08 (m, 4H). HPLC 99.7%
PAR; >99.5% ee. Method (related substances): Spherisorb
ODS II column, 3 µm, 125 × 4.6 mm, injection volume 10
µL, mobile phase 7:3 [aqueous acetic acid (0.58 M) and
sodium acetate (0.01 M)]:[acetonitrile/2-heptylamine 60:1],
flow rate 0.8-0.9 mL/min, detection at 278 nm, run time
40 min, retention times 1 6.2 min; 12 38.3 min. Method
(ee): Diacel Chiralpak AD column, 10 µm, 250 × 4.6 mm,
injection volume 20 µL [1.0 mg/1.0 mL], mobile phase 9:1
heptane: 0.1% diethylamine in 2-propanol, flow rate 1.0 mL/
min, temperature 20-25 °C, detection at 230 nm, run time
20 min; retention times 1b, 8.6 min; 1a, 10.0 min.
(R)-Verapamil (1a). 1a was prepared from 12a by the
same method in 60.7% overall physical yield,22 following
2-propanol recrystallisation of the hydrochloride salt (mp
132-134 °C, [R]25D ) +8.9° (c 5.0, ethanol), [lit.7 mp 131-
133 °C; [R]24D ) +8.9° (c 5.01, ethanol]). 1H NMR (D2O):
identical spectrum to 1b; HPLC purity 99.8% PAR; >99.5%
ee.
mol, 85.5% yield; mp 95-97 °C), [R]25 ) -6.1° (c 5.18,
D
ethanol).
1H NMR (d6-Me2SO, at 20 °C): δ 0.71 (2d, 3H); 1.14
(2d, 3H); 1.70 (m, 0.5H); 1.87 (m, 0.5H); 1.99 (septet, 0.5H);
2.1-2.32 (m, 3.5H); 2.60 (2t, 2H); 2.73 (s, 1.5H); 2.80 (s,
1.5H); 3.3 (t, 1H); 3.42 (t, 1H); 3.7-3.8 (4s, 12H); 6.50 (d,
0.5H); 6.67 (s, 0.5H); 6.68 (d, 0.5H); 6.77 (s, 0.5H); 6.78
(d, 0.5H); 6.84 (d, 0.5H); 6.91-7.02 (m, 3H). (at 100 °C):
δ 0.77 (d, 3H); 1.15 (d, 3H); 1.92 (m, 1H); 2.18-2.36 (m,
4H); 2.67 (t, 2H); 2.76 (s, 3H); 3.4 (t, 2H); 3.76-3.81 (4s,
12H); 6.67 (d, 1H); 6.76 (s, 1H); 6.86 (d, 1H); 6.98 (m, 3H).
HPLC: 99.0% PAR; Method: Novapak C18 column; mobile
phase water 60: acetonitrile 40: TFA 1; 1 mL per min,
detection at 230 nm. Retention times: 11, 1.5 min; 2, 3.9
min; 12, 9.7 min. Calcd for C27H36N2O5: C, 69.21; H, 7.74;
N, 5.98%. Found: C, 69.3; H, 7.75; N, 5.86%. Mass
spectrum (m/z) 469.3 (M + 1)+.
(R)-Verapamilamide (12a). 12a was prepared by the
same method from 10a and 11 in 82.3% yield (mp 94.8-
1
97.1 °C, [R]25 ) +6.0° (c 4.9 ethanol). The H NMR
D
spectrum was identical to that of 12b and LC purity was
98.8% PAR. CHN Found C, 69.47; H, 7.69; N, 5.79%.
Racemic 12 was also made from 2 in a similar fashion in
88.6% yield and had mp 105-106.5 °C with LC purity
99.5% PAR.
(S)-Verapamil (1b). Amide 12b (100 g, 0.214 mol) was
suspended in THF (450 mL) and cooled to 0-5 °C. BMS
(30.0 g, 38 mL, 0.395 mol) was then added dropwise over
20 min at this temperature. After a THF line wash (2 × 25
mL), the mixture was stirred out for 2 h before being allowed
Acknowledgment
We thank Professor Andrew Pelter and Dr. John O’Rouke
for helpful discussions concerning the amide reduction stage,
as well as Dr. Steven Lloyd for advice on HPLC method
development. The analytical support provided by Ms. Cathe-
rine Rippe, Mrs. Christel Kronig, Mr. Philip Gilbert, Dr.
Rawl Hardial, Mr. Nicolas Dawe, and Ms. Barbara Mason
is also greatly appreciated.
Received for review June 1, 2000.
OP000059Q
(22) The physical yields quoted are net of the quantity of authentic seed added.
472
•
Vol. 4, No. 6, 2000 / Organic Process Research & Development