Synthesis of the quinolone ABT-492: Crystallizations for optimal processing

Article abstract of DOI:10.1021/op060054e

ABT-492 has been under development at Abbott Laboratories as a quinolone antibiotic. A convergent syntheses was utilized to prepare the compound on a multi-kilogram scale. Difficulties in isolation of intermediates were overcome by developing control of t

Full text of DOI:10.1021/op060054e

Organic Process Research & Development 2006, 10, 751−756
Synthesis of the Quinolone ABT-492: Crystallizations for Optimal Processing
Anthony R. Haight,* Sema Z. Ariman, David M. Barnes, Nancy J. Benz, Francoix X. Gueffier, Rodger F. Henry,
Margaret C. Hsu, Elaine C. Lee, Larry Morin, Kurt B. Pearl, Matthew J. Peterson, Daniel J. Plata, and David R. Willcox
GPRD Process Research and DeVelopment, Abbott Laboratories, Bldg. R8/1, 1401 Sheridan Road,
North Chicago, Illinois 60064-6285, U.S.A.
Abstract:
Following chlorination of 7b, the ester groups were
hydrolyzed and the free acid 9 was isolated in 91% yield
over the two steps (Scheme 2). To improve the solubility
and bioavailability of the quinolone, a N-methyl-d-glucamine
ABT-492 has been under development at Abbott Laboratories
as a quinolone antibiotic. A convergent syntheses was utilized
to prepare the compound on a multi-kilogram scale. Difficulties
in isolation of intermediates were overcome by developing
control of the physical forms. Examples of controlling the
agglomeration, crystal habit, and polymorphism of intermedi-
ates and the API are described.
5
salt of 9 was prepared. The efficiency and brevity of this
sequence encouraged us to utilize this approach to provide
kilogram quantities of ABT-492 for clinical development.
To improve the scaleability, we focused our attention on
controlling the physical properties of the intermediates for
handling purposes.
Introduction
While crystallization from solution is an established
technique for the separation and purification of organic
compounds, challenges due to complex interactions of
variables including polymorphism, crystal nucleation and
growth, and impurity profile are common. For drug sub-
stances, understanding and controlling the polymorphism,
particle size, morphology, and crystal habit have become a
Vinyl Amide, 4
Vinyl amide 4 was chosen as the first isolated intermediate
since the subsequent base induced cyclization necessitated
removal of acetic acid, a byproduct of the vinyl amide
formation. Additionally, residual diaminopyridine was shown
to act as a competitive nucleophile with 6, further supporting
4
the decision to isolate 4.
1
fundamental steps in pharmaceutical development programs.
Ketoester 1 was condensed with triethylorthoformate
Less interest, however, has been focused on the manipulation
of the properties of pharmaceutical intermediates to improve
(TEOF) in the presence of acetic anhydride. The resulting
ethoxymethylene ester, 2, was not isolated but was instead
diluted with an acetonitrile/NMP mixture and treated with
sufficient water to consume residual TEOF. Excess water
led to partial hydrolysis of 2 resulting in formation of 4-oxo-
benzopyran 11 under the cyclization conditions (eq 1). The
2
ease of handling or purification. This is surprising in view
of the impact physical forms have on purification and
mechanical isolation.
6
3
In conjunction with a clinical program, we required
kilogram quantities of the quinolone antibiotic ABT-492. For
this, we developed a scaleable synthetic approach highlighted
4
by a selective chlorination methodology. The synthetic
approach to ABT-492 entails condensation of triethylortho-
formate with ketoester 1, to give, after coupling with 2,6-
diamino-3,5-difluoropyridine 3, the vinylamide 4 in >95%
isolated yield (Scheme 1). Cyclization to intermediate 6 was
accomplished by treating a LiCl complex of 4 with DBU.
In situ coupling of 6 with azetidinol 5 in the presence of
additional DBU yielded, after esterification, the isobutyrate
ester 7b in 93% isolated yield.
*
To whom correspondence should be addressed. E-mail: anthony.haight@
abbott.com.
ethoxymethylene ester was added to a solution of 2,6-
diamino-3,5-difluoropyridine 3 in NMP/acetonitrile giving
a yellow solution of the desired amide 4. Slow addition of
water gave bright yellow needles of 4 (Figure 1a) in >93%
isolated yield. Unfortunately, these crystals filtered excep-
tionally poorly leading to difficulties in removing both
reaction byproducts and residual solvents. The product
(
1) (a) Baskin, R. J.; Bowker, M. J.; Slater, B. J. Org. Process Res. DeV. 2000,
4
, 427. Myerson, A. S., Ed. Handbook of Industrial Crystallization;
Butterworth-Heinemann: Woburn, MA, 1992. (b) Desikan, S.; Anderson,
S. R.; Meenan, P. A.; Toma, P. H. Curr. Opin. Drug DiscoVery DeV. 2000,
3
, 723.
(
2) Mueller, M.; Schneeberger, R.; Wieckhusen, D.; Cooper, M. Org. Process
Res. DeV. 2004, 8, 376.
(
3) (a) Harnett, S. J.; Fraise, A. P.; Andrews, J. M.; Jevons, G.; Brenwald, N.
P.; Wise, R. Journal of Antimicrobial Chemotherapy 2004, 53, 783-792.
(b) Da Silva, A. D.; de Almeida, M. V.; de Souza, M. V. N.; Couri, M. R.
C. Current Medicinal Chemistry 2003, 10, 21-39.
(5) Berge, S. M.; Bighley, L. D.; Monkhouse, D. C. J. Pharm. Sci. 1977, 66,
1.
(6) If the excess TEOF was not hydrolyzed, N-formylation impurities of 3 and
4 were observed in the reaction mixture.
(
4) Barnes, D. M.; Christensen, A.; Engstrom, K. M.; Haight, A. R.; Hsu, M.
C.; Lee, E. C.; Peterson, M.; Plata, D. J.; Raje, P.; Stoner, E. J.; Tedrow, J.
S.; Wagaw, S. H. Org. Process Res. DeV. 2006, submitted.
1
0.1021/op060054e CCC: $33.50 © 2006 American Chemical Society
Vol. 10, No. 4, 2006 / Organic Process Research & Development
•
751
Published on Web 06/23/2006

Scheme 1. Synthesis of ABT-492 intermediate
Scheme 2. Synthesis of ABT-492
obtained from a 40 kg run after drying under a vacuum to
remove residual moisture contained approximately 5 mol %
NMP and had a weight percent assay of 98.1% versus a
chromatographically prepared standard
ably the rapid decrease in solubility caused by the inverse
quench crystallization induces agglomeration during the
dispersion of the organic phase into the water. X-ray
diffraction patterns of the agglomerated crystals indicated
no form changes between the needles and the agglomerates,
nor were the impurity profiles of the solids significantly
different between the methods of precipitation, 95.4-96.9%
versus 96.9-98.1%.
8
Modulating the morphology by crystallizing from alterna-
tive solvents or mixed solvents proved unsuccessful. Ostwald
ripening (60 °C/12 h) did increase the needle size; however,
7
no improvement was observed in the cake resistivity.
Additionally, Ostwald digestion resulted in higher levels of
impurities including bis-amide 12 (<4% to 5-15%) and
N-acetamide 13 (<1% to >5%). N-Acylation of 4 from
residual Ac O results in acetamide 13, while 12 is the product
2
of metathesis of 4 under the digestion conditions.
By inverting the order of addition and adding the reaction
mixture to water, soft agglomerates of fine needles (100-
In practice, the inverse addition precipitation on a multi-
kilogram scale using a retreating blade agitator produced
acceptable agglomerates. Yet during the mixing prior to
filtration of the solids (∼1 h), the agglomerates were
observed to undergo significant degradation to fine needles
leading again to poor filtration and cake washing.9
(8) Kawashima, Y.; Okumura, M.; Takenaka, H. Science 1982, 216 (4550),
127.
800 um agglomerates) (Figure 1b) were observed. Presum-
1
(
9) Laser diffraction analysis of a 200 gallon reactor equipped with a pitched
blade agitator with a tip speed of 14.4 m/min showed an attrition in the d90
of the chord length from 95 µm to 70 µm over 3 h.
(7) Particle size was measured by microscopy, as laser diffraction techniques
proved unsuccessful in differentiating particles.
752
•
Vol. 10, No. 4, 2006 / Organic Process Research & Development

attrition relative to the retreat curve.12 On a 10 kg scale, we
observed little to no attrition of the agglomerates employing
a pitched blade agitator. Scaling the precipitation based on
13
the power density (W/L), agglomerates of 4 were produced
on greater than a 50 kg scale with minimal attrition resulting
in reproducible filtration rates (0.8 to 1.0 g/s) and perme-
-14
2
abilities ((1.3 to 2.6) × 10 m ) to provide a more preferred
isolation process. Filtration times on this scale decreased from
2
4 h to 90 min, and the product was isolated in 91-95%
isolated yield and greater than 95% purity on a 50-125 kg
scale (8 runs).
Isobutyrate Ester, 7b
Complexation of vinyl amide 4 with LiCl in NMP
followed by treatment with 1 equiv of DBU initiated an
anionic cyclization to the quinolone core, 6. The 3-hy-
droxyazetidinyl functionality was prepared in situ following
incremental addition of 3-hydroxyazetidine HCl and 2 equiv
of DBU to the mixture. We found it advantageous to esterify
the resulting alcohol 7a to improve the solubility character-
4
istics for the subsequent chlorination step.
By following the three-step sequence of cyclization,
coupling, and protection, quinolone ester 7b was precipitated
by addition of aqueous citric acid. Isolation at this point was
necessary to remove NMP and residual azetidinol prior to
chlorination.
Three distinct crystalline forms of the isobutyrate ester
7
b were characterized (Figure 2). The first characterized
material was isolated following chromatographic purification.
XPRD analysis of this lot (vide infra) showed Form I
contaminated with Form II (Figure 2). Form I has not been
observed since. Forms II and III have been observed
repeatedly in essentially pure states during crystallization of
the reaction mixture (Figure 2).
Form II is usually associated with a “starburst” crystal
habit (Figure 3a) and was initially isolated by a slow addition
of water in a drowning crystallization. While Form II
particles filter well immediately after completion of the water
addition, the starburst habit did not prove robust to agitation
necessary to suspend the solids prior to isolation. As a result,
a significant amount of attrition occurred during agitation
resulting in a poor separation of the solid and liquid phases
as evidenced by variable levels of lithium (0.1-2.0%),
Figure 1. Vinyl amide 4 quench morphologies.
Comparison of filtration rates and permeabilities demon-
strated that agitation during the inverse quench was a
fundamental parameter in influencing the filtration rate (Table
1
, entries 2-4).10 However, the resulting agglomerates
underwent competitive destruction with a prolonged mix time
(Table 1, entries 4-7). This encouraged us to minimize the
time for completing the addition and the holding of the slurry
prior to isolation.11
Although no difference was observed between a pitched
blade agitator and a retreat curve design in the particle
formation, the pitched blade agitator did reduce the rate of
Table 1. Vinyl amide 4 isolation
hold
time
(h)
filtration
rate
(g/s)
a
addition
time (h)
agitator
blade
agitation
(W/L)
permeability
2
-14
entry
(m × 10
)
1b
2
3
4
5
6
7
8
9
retreat
retreat
1.64 × 10
-3
2
0
0
0
0.5
0
0
0
7.5
0.21
0.54
0.91
1.27
0.86
1.54
0.53
1.07
0.83
Nm
-
4
1
1
1
1
3
6
2.1 × 10
0.52
2.12
4.34
2.99
1.42
0.88
2.61
1.30
-
4
pitched
pitched
pitched
pitched
pitched
pitched
pitched
4.1 × 10
-
-
3
3
3.26 × 10
3.26 × 10
3.26 × 10
3.26 × 10
3.19 × 10
3.19 × 10
c
-3
-
-
-
3
3
3
3.1
3.1
d
a
3
2
2
Calculated using Darcy’s equation: (Q/A) ) K * (∆P/L) * (g
c
/µ), where Q ) volumetric flow rate (m /s); A ) filter surface (m ); K ) permeability constant (m );
b
∆
H
P ) pressure drop across filter (Pa); L ) depth of filtercake (m); g
O/mol of 1 versus nominal (2 kg of H
) gravitational constant; µ ) liquid viscosity (Pa ‚ s). Water added to reaction. c 4 kg of
c
d
-4
2
2
O/mol of 1). Power density held at 2.1 × 10 W/L during hold time.
Vol. 10, No. 4, 2006 / Organic Process Research & Development
•
753

Figure 2. XRPD of isobutyrate 7b isomers.
Figure 4. Isobutyrate 7b quench.
tion. In addition, the individual levels of lithium, chloride,
and fluoride in the isolated solids are typically less than 200
14
ppm. The filterability and washing of the Form III particles
encouraged us to assess reproducible generation of this
morphology.
Conversion to Form III was observed in mixtures of
Forms II and III in a variety of solvents including partially
quenched reaction mixtures. If the partially quenched reac-
tions had supernatant levels greater than 50 mg of 7b/mL of
supernatant, the conversion was observed to occur in less
1
5
than 1 h.
Thus, the reaction mixture was diluted sequentially with
1
6
ethyl acetate (1 mL/g of 3) and 10 wt % aqueous citric
acid (1 g/g of 3) to generate a solution with an approximate
solution concentration of 70 mg of 7b/mL of solution. Form
III seeds (1 wt %) were added as a slurry in NMP/aqueous
1
7
citric acid, and the resulting mixture was allowed to
desaturate until the supernatant levels reached equilibrium
(approximately 65-69 mg/mL; Figure 4). Additional aque-
Figure 3. Isobutyrate 7b polymorph habits.
1
8
ous citric acid was then added slowly.
chloride (0.1-2.6%), and fluoride (0.1-0.5%) in the isolated
solids. Presumably the high cake resistivity inhibits effective
removal of the liquors (vide infra).
Form III is typically isolated as plates by addition of 10%
aqueous citric acid to the reaction mixture (Figure 3b). These
plates proved more mechanically stable to prolonged agita-
(
10) Agitation was increased by increasing the rpm of the agitator blade in a
vessel. Scaling up based on the power density (W/L) appeared to give
comparable agglomerates on several scales.
(11) Addition rates between 1 and 3 h were reproducibly obtained on scale with
hold times of 2-5 hours
(12) Impellers such as the A-310 used here are designed to improve axial flow
and mixing with reduced power input.
754
•
Vol. 10, No. 4, 2006 / Organic Process Research & Development

Figure 5. Neutralization of hydrolysis reaction.
Following washing with water to remove residual inor-
ganics, isobutyric acid, DBU, and NMP, the isobutyrate ester
7b was isolated as large plates in 96% yield and >99% purity
with less than 50 ppm of Li, Cl, and F.
Free Acid Isolation
With the isobutyrate ester 7b prepared, introduction of
the chlorine atom and hydrolysis of the protecting groups
were all that remained. To a solution of NCS in methyl
acetate is added an ethyl acetate solution of 7b containing
2 4
6.5 mol % H SO . The MeOAc/EtOAc mixed solvent system
was developed to ensure the starting ester 7b, and the
resulting chlorinated quinolone 8 stayed in solution through-
4
out the chlorination sequence. The resulting reaction is
washed sequentially with aqueous sodium bicarbonate fol-
lowed by aqueous sodium bisulfite. After solvent exchange
into 2-propanol and addition of excess aqueous NaOH, the
reaction mixture is heated to 50 °C to complete the hydrolysis
of the esters. The isobutyrate cleaves rapidly, while the ethyl
ester requires the elevated temperature to ensure a rapid rate
of hydrolysis. The product is isolated by a reactive crystal-
lization in which 12.4 wt % acetic acid is slowly added to
the reaction mixture. A large excess of acetic acid is added
at this point to drive the pH of the reaction mixture to less
than 5.0 and precipitate the product. The solubility of ABT-
Figure 6. ABT-492 crystal habits (micrographs same scale).
The driving force for the precipitation is so great between
the addition of the first and second equivalents of acid (35
mg/mL to <5 mg/mL) that the pH increases, from 7.4 to
7
.7, during the crystallization.
492 increases significantly between pH 8.5 and 10.0, and
After completion of the crystallization, the slurry is stirred
thus, the rate of addition of the acid was controlled in this
range (Figure 5).19
at 40 °C for at least 30 min to ensure that entrained potassium
salt in the solid has sufficient time to equilibrate. The slurry
is then cooled to 20 °C and filtered.
Washing of the cake is critical to removing both potassium
acetate and acetic acid. To ensure removal of the acetic acid,
the pH of the filtrate is monitored until the eluent stream
matches the pH of the water used for the wash. After drying,
the ivory solid, 9, is obtained in 90% assay adjusted yield
with an assay purity of >95%.
The first equivalent of acetic acid is added over a 5-15
min period. The contents of the reactor are then stirred at
low agitation20 for 1 h during which time the solution
becomes opaque with the initial crystals. The agitation is
kept only high enough to suspend the solids forming in the
vessel and to maintain a homogeneous mixture. The second
equivalent of the 12.4 wt % acetic acid is added over 1 h
followed by an additional 3 equiv of acid over another hour.
Unfortunately, the poor solubility characteristics of the
acid 9 precluded evaluation of this compound in a clinical
setting. A salt screen of pharmaceutically relevant counter-
(
(
(
13) Bates, R. Mixing Theory and Practice; Academic: New York; Vol. 1,
Chapter 3, p 111.
14) This proved critical for the following NCS chlorination step which appeared
to be catalyzed by Li and conversely inhibited by fluoride.
15) The solution concentration of 7b is the critical in the rate of interconversion
between Forms II and III. Solution concentrations <20 mg/mL can require
days for complete conversion.
6
ions identified the N-methyl-d-glucamine salt, ABT-492, as
a viable candidate capable of providing bioavailability. Initial
quantities of ABT-492 were acquired by crystallization from
1
0 to 30% aqueous 2-propanol to yield large needles of the
(
(
(
(
(
16) EtOAc was added to decrease the viscosity of the mixture to improve the
crystal growth.
trihydrate of the desired salt in >98% purity (Figure 6a).
The anhydrous form of the compound could then be obtained
via dehydration under a vacuum.
This needle morphology, while acceptable for early
clinical evaluation of the compound, proved challenging to
formulate due to the poor flow characteristics of the solid.
17) For ease of operation, the seeds were added in a slurry. Lab experiments
with dry seeding also was successful.
18) Slow addition was critical to avoid formation of Form II which does not
convert in an acceptable time frame to Form III after the quench.
19) Solubilities: pH 8.5 ) 4 mg/mL; 9.0 ) 16 mg/mL; 9.5 ) 55 mg/mL; 10.0
)
194 mg/mL
20) A retreating bar agitator was used with a tip speed of 9.6 m/min.
Vol. 10, No. 4, 2006 / Organic Process Research & Development
•
755

Bravais-Friedel-Donnay-Harker modeling21 of the trihy-
drate crystal structure suggested an oblong hexagonal lathe
should be the preferred habit. Surveying of crystallization
solvents and methods determined that crystallization directly
from water generated a comparable plate habit (Figure 6b)
presumably due to the increased water concentration inhibit-
ing growth of the other faces.22
In practice, seeding an aqueous mixture of 9 and 1.2 equiv
of N-methyl-d-glucamine at 38 °C, followed by slow cooling
yielded, after filtration at 0 °C, 80% of ABT-492 in >99%
purity.
Thus, we have described examples of successful applica-
tions of crystallization techniques to improve both isolations
and purifications in the development of a pharmaceutical
process. The utilization of crystal form and habit to improve
solid/liquid separation and thus enhance purity should be
generally considered when confronted with scaling of
pharmaceutical processes.
5 (33.9 kg, 309 mol, 1.08 equiv), is added followed by DBU
(109.2 kg, 717 mol, 2.5 equiv) over 2 h. The reaction
temperature is then adjusted to 23 ( 5 °C. After the addition
is complete, isobutyric anhydride (99.7 kg, 630 mol, 2.2
equiv) is added, the reaction is stirred for 1 h at 35 °C and
then cooled to 23 °C, and ethyl acetate (104.2 kg) is added.
A 10% citric acid solution (300 kg) is added slowly. Seeds
of 7b are added. An additional 270 kg of 10% citric acid
solution are then added, and the resulting slurry is filtered
and washed twice with 115 kg of water. The product is dried
at 50 °C to yield 136.1 Kg (93%) of the title product, 7b.
1
mp: 201-209 °C; H NMR (DMSO-d
6
, 400 MHz) δ 8.49
(s, 1H), 8.00 (dd, J ) 9.0, 9.3 Hz, 1H), 7.75 (d, J ) 12.8
Hz, 1H), 6.79 (br s, 2H), 5.95 (dd, J ) 1.5, 7.6 Hz, 1H),
5.21 (m, 1H), 4.36 (t, J ) 7.4 Hz, 2H), 4.02 (q, J ) 7.0 Hz,
2H), 3.95 (dd, J ) 3.7, 9.2 Hz, 2H), 2.58 (hept, J ) 7.0 Hz,
1H), 1.26 (t, J ) 7.0 Hz, 3H), 1.11 (d, J ) 7.0 Hz, 6H).
Anal. Calcd for C24
Found: C, 56.86; H, 4.30; N, 10.89.
-(6-Amino-3,5-difluoropyridin-2-yl)-8-chloro-6-fluoro-
-(3-hydroxyazetidin-1-yl)-4-oxo-1,4-dihydroquinoline-3-
23 3 4 5
H F N O : C, 57.14; H, 4.60; N, 11.11.
Experimental Section
1
3-(2,4,5-Trifluorophenyl)-2-(N-[6′-amino-3′,5′-difluoro]-
7
aminopyridine)methylene-3-oxopropionic Acid, Ethyl Es-
ter, 4. Ketoester 1 (83.2 kg, 337.9 mol) and triethylortho-
formate (80.1 kg, 540.7 mol) are heated to reflux (∼140 °C)
and stirred for 30 min. Acetic anhydride (103.5 kg, 1014
mol) is then added, and heating continued for about 12 h,
and then the mixture was cooled and diluted with NMP (210
kg) and acetonitrile (161 kg) and stirred. Water (3.0 kg, 169
mol) is added, and the mixture was held for 10 min.
The resulting solution is added to a suspension of 2,6-
diamino-3,5-difluoropyridine, 3 (57.4 kg, 395.4 mol), NMP
carboxylic Acid, 9. A solution of N-chlorosuccinimide (25.3
kg) in methyl acetate (419 kg) at 17 °C was treated with
sulfuric acid (0.56 kg, 5.7 mol). This solution was transferred
to a slurry of isobutyrate ester 7b (92.7 kg, 184 mol) in ethyl
acetate (244 kg) at 17 °C while maintaining the reaction
temperature at 17 °C. The solution was stirred at 17 °C for
1
-8 h and diluted with 1.5% (w/w) NaHCO
layers were separated, and the organics washed with 10%
w/w) aqueous sodium sulfite (200 kg) and concentrated. The
concentrate was dissolved in 2-propanol, treated with 4%
w/w) KOH (750 kg), and stirred at 50 °C until hydrolysis
3
(370 kg). The
(
(210 kg), and acetonitrile (161 kg). The resulting homoge-
(
neous solution is added to water (662 kg) over 2 h,
precipitating the yellow product. The product is filtered and
washed with a solution of acetonitrile (161 kg) and water
was complete. The reaction was passed through a filter,
treated with 12% (w/w) acetic acid (410 kg), and filtered.
The filtrate was washed with water and dried at 50 °C to
(102 kg). The wetcake is washed with water (600 kg) and
1
provide 73 kg (90%) of product 9. Mp: 238-241 °C. H
dried at 60 °C to provide 120 kg (93%) of the vinyl amide
NMR (CDCl
7
6
3
) δ 14.63 (brs, 1H), 8.70 (d, J ) 0.7 Hz, 1H),
.95 (dd, J ) 9.9, 0.7 Hz, 1H), 7.83 (d, J ) 13.6 Hz, 1H),
.75 (s, 2H), 5.75 (d, J ) 5.8 Hz, 1H), 4.61 (m, 12H), 4.47
: C,
1
4
, as a mixture of E and Z isomers. Mp 157-160 °C; H
NMR (CDCl , 300 MHz) (E-isomer) δ 1.15 (t, 3H), 4.16
q, 2H), 4.64 (br s, 2H), 6.90 (m, 1H), 7.22 (t, 1H), 7.32 (m,
3
(
(m, 1H), 4.18 (m, 2H). Anal. Calcd for C18
H12ClF
N O
3 4 4
1H), 9.03 (d, 1H), 12.44 (bd, 1H); (Z-isomer) δ 1.03 (t, 3H),
4.11 (q, 2H), 4.60 (br s, 2H), 6.90 (m, 1H), 7.20 (t, 1H),
7.48 (m, 1H), 8.90 (d, 1H), 11.17 (bd, 1H). Anal. Calcd for
49.05; H, 2.74; N, 12.71. Found: C, 48.90; H, 2.48; N, 12.62.
N-Methyl-d-glucammonium 1-(6-Amino-3,5-difluoro-
pyridin-2-yl)-8-chloro-6-fluoro-7-(3-hydroxyazetidin-1-
yl)-4-oxo-1,4-dihydroquinoline-3-carboxylate, ABT-492.
To a vessel containing acid 9 (29.6 kg, 67.2 mol) and
N-methyl-d-glucamine (18.4 kg, 94.3 mol) was added
distilled water (133 kg). This was heated to 60 °C to dissolve
the solids and then cooled to 38 °C. An aqueous slurry of
seeds was added to the mixture (0.64 kg, 1.5 wt %), and the
solution was held at 38 °C for 24 h before cooling
parabolically to 0 °C. The slurry was filtered and washed
with 25% (w/w) aqueous 2-propanol. The solid was dried at
C
17
H
12
F
5
N
3
O
3
: C, 50.88; H, 3.01; N, 10.47. Found: C, 50.83,
H, 2.70, N, 10.32.
-(6-Amino-3,5-difluoropyridin-2-yl)-6-fluoro-7-(3-isobu-
1
tyryloxyazetidin-1-yl)-4-oxo-1,4-dihydroquinoline-3-car-
boxylic Acid, Ethyl Ester, 7b. To a solution of vinyl amide
4
(115.5 kg, 288 mol) and LiCl (24.3 kg, 573 mol, 1.99
equiv) in NMP (769 kg) is added DBU (46.1 kg, 303 mol,
.06 equiv) over 1 h 40 min, maintaining the internal
1
temperature at 35 °C. The reaction temperature is then
adjusted to 23 ( 5 °C, and the reaction is stirred for 2 h.
When the cyclization is complete, azetidine hydrochloride,
3
1
0 °C to provide 31.7 kg of product (72.9%). Mp: 168-
71 °C.
(21) Docherty, R.; Clydesdale, G.; Roberts, K. J.; Bennema, P. J. Phys. D: Appl.
Phys. 1991, 24, 89.
(
22) Even in the presence of seeds, the desupersaturation rate requires ap-
proximately 48 h at ∼38 °C to reach equilibrium in water versus <2 h in
Received for review March 10, 2006.
OP060054E
2
0% aqueous 2-propanol.
756
•
Vol. 10, No. 4, 2006 / Organic Process Research & Development