Kilogram-scale synthesis of a highly selective α1-adrenoceptor antagonist (DL-028A)

Article abstract of DOI:10.1021/op0100807

This work presents an improved eight-step process, leading to kilogram quantities of high-quality DL-028A, an antihypertensive agent. The improvements include reducing the levels of toxic reagents and the removal of dangerous processes and waste gas treatment. Moreover, specification and impurity profiles were determined.

Full text of DOI:10.1021/op0100807

Organic Process Research & Development 2002, 6, 273−278
Kilogram-Scale Synthesis of a Highly Selective r -Adrenoceptor Antagonist
1
(DL-028A)
Shan-Yen Chou,* Wei-Kung Yin, Yuh-Shan Chung, Lien-Shange Chang, Chin-Wei Liu, Shyh-Fong Chen, and
Kae-Shyang Shih
Chemical DeVelopment DiVision, DeVelopment Center For Biotechnology, 102, Lane 169, Kang Ning Street, Hsichih,
Taipei Hsien, Taiwan, R. O. C.
Abstract:
This work presents an improved eight-step process, leading to
kilogram quantities of high-quality DL-028A, an antihyperten-
sive agent. The improvements include reducing the levels of
toxic reagents and the removal of dangerous processes and
waste gas treatment. Moreover, specification and impurity
profiles were determined.
Figure 1.
Introduction
Scheme 1. Original methoda
10,11
Recent research in the field of R blockers has led to the
discovery of several biologically specific R-adrenoceptor
1-4
antagonists. More recently, some R
1
-antagonists are being
investigated for treating dysuria secondary to benign prostatic
5
,6
7
hypertrophy. DL-028, one newly developed R
1
-antagonist,
8
-10
is highly selective and with LDL-lowering properties.
Furthermore, according to an in vivo antihypertensive
experiment, the cardiotoxicity is greatly reduced by combi-
nation infusion with prazosin.10 Consequently, DL-028 was
selected as a leading molecule by the Development Center
For Biotechnology (DCB) for preclinical study.
a
2 3 2
a. CS /pyridine, reflux; b. CH I/NaOH, H O; c. allylamine/MeCN, autoclave,
1
20-130 °C; d. NBS/MeCN.
Chern et al. (1993) described the original synthesis of DL-
0,11
0
28 for the first time.1
The linear pathway employed a
chemistry therein from bench scale to kilogram scale.
However, the disclosed method required modifications to
provide a scaleable process and to eliminate potentially
hazardous reactions. This work concerns the preparation of
kilogram quantities of DL-028 hydrochloride (i.e., DL-028A)
to preclinical and toxicological specification, since a DL-
028 salt had not been prepared and evaluated before.
fused quinazoline 5 as the substrate that was coupled with
the commercially available phenylpiperazine to generate the
DL-028 skeleton (Figure 1). The same investigators later
described an asymmetric synthesis of DL-028 in homochiral
form, in which some modification on the synthesis of early
quinazoline intermediates had been reported. This work is
concerned with part of their study and intends to extend the
12
Results and Discussion
(1) Timmermans, B. M. W. M.; van Zweieten, P. A. Drugs Future 1984, 9,
4
1.
Reevaluation of the Original Process (Scheme 1).10,11
The process reevaluation for the key intermediate 5 is
(
(
(
2) Imagawa, J.; Sakai, K. Eur. J. Pharmacol. 1986, 131, 257.
3) Fozard, J. R. J. CardioVasc. 1982, 4, 829.
4) Cohen, M. L.; Fuller, R. W.; Kurz, K. D. Hypertension (Dallas) 1983, 5,
described as follows (Scheme 1). In step one of the original
6
76.
10,11
process,
the quinazolinedithione 2 was formed by re-
(
(
5) Koga, T.; Shiraki, Y.; Sakai, K. Jpn. J. Pharmacol. 1989, 50, 185.
6) Seki, N.; Nishiye, E.; Itoh, T.; Susuki, H.; Kuriyama, H. Br. J. Pharmacol.
fluxing 2-aminobenzonitrile 1 with excess carbon disulfide
in pyridine. However, carbon disulfide is a poisonous and
highly flammable compound; moreover, the irritant pyridine
is not amenable to scaling-up because of difficulties in
workup. In step two of the original process, methylation of
2 required 2 mol equiv of methyl iodide, which is a volatile
toxic and expensive agent. Consequently, this process is
costly and inconvenient to perform. In step three of the
original process, the substitution reaction was performed in
an autoclave and required excessive amounts of highly
volatile allylamine to displace the methylthio group regio-
1
988, 93, 702.
7) Receptor binding studies revealed that the affinity constant (K
for R -adrenoceptor is more than 2000-fold greater than that for R
(
(
(
i
) of DL-028
receptors.
1
2
8) Peng, I.-H.; Sheu, J.-R.; Chern, J.-W.; Lee, Y.-M.; Yen, M.-H. J. Biomed
Sci. 1996, 3, 108.
9) Huang, Y.-T.; Lin, L.-C.; Chern, J.-W.; Lin, H.-C.; Hong, C.-Y. Pharmacol-
ogy 1998, 57, 132.
10) Chern, J.-W.; Tao, P.-L.; Yen, M.-H.; Lu, G.-Y.; Shiau, C.-Y.; Lai, Y.-J.;
Chien, S.-L.; Chan, C.-H. J. Med. Chem. 1993, 36, 2196 and references
therein.
(
(
11) Chern, J.-W.; Lu, G.-Y.; Lai, Y.-J.; Yen, M.-H.; Tao, P.-L. U.S. Patent
5
,340,814, 1994.
12) Gutcait, A.; Wang, K.-C, Liu, H.-W.; Chern, J.-W. Tetrahedron: Asymmetry
996, 7, 1641.
(
1
1
0.1021/op0100807 CCC: $22.00 © 2002 American Chemical Society
Vol. 6, No. 3, 2002 / Organic Process Research & Development
•
273
Published on Web 04/17/2002

Scheme 2. Modified method (for kilogram scale-up
Scheme 3. Synthesis of DL-028Aa
production)a
a
a. CS
2
, KOH/EtOH, reflux, 99.3%; b. CH
3 2 4 2
) SO /NaOH, H O, 82.5%; c.
3
POCl /ClCH
2
CH Cl, NEt ; d. allylamine/THF, reflux, 50.9%, two steps (8-4);
2
3
e. NBS/MeCN, 1.5 h, 82.0%.
selectively. The reaction is not suited for scaling-up because
of serious air pollution and a potentially dangerous high-
pressure operation. In step four of the original process, the
bromination of 4 with concomitant dihydroimidazole ring
formation was performed out in acetonitrile with N-bromo-
succinimide. However, the bromide 5 is unstable under
prolonged stirring (∼3 h), and several unidentified side
products were observed; the extent of the reaction required
frequent analysis, and the reaction period was minimized.
Accordingly, a modified approach to these compounds was
sought.
Kilogram Processes of the Intermediate 5 (Scheme 2).
A modified reaction sequence was employed for kilogram-
scale synthesis to circumvent problems associated with safety
and cost and the concern of scale-up feasibility. The sequence
employs anthranilamide 6, which is much less expensive than
a
3
a. NaHCO (in portions)/MeCN, reflux, 66.0%; b. 1) NaOH/aqueous
methanol, reflux; 2) aqueous sodium periodate, 35 °C, 81.0%; c. concd HCl/
MeOH, 83.4%.
than those of original process.10,11 The isolated yield of 4 is
0.9% (8-4, two steps). In the fifth step, bromination with
5
concomitant cyclization was performed in the presence of
N-bromosuccinimide (∼120 mol % vs 4) in acetonitrile. The
extent of reaction was checked by HPLC every 15 min. On
completion of the reaction, the reaction was immediately
cooled to avoid decomposition and to precipitate the bro-
minated product 5, which was obtained as a white powder
in 82.0% isolated yield on filtration. In one laboratory-scale
experiment, 0.5 mol equiv of 1,3-dibromo-5,5-dimethyl-
hydantoin was employed as the bromination agent, yielding
only 54.5%. This was not developed further even though
the atom efficiency is higher than that of NBS, theoretically.
Unlike the original method, the newly developed five-step
sequence is safe, economical, and suited to scale-up, although
one more step is required.
2
-aminobenzonitrile 1, as the starting material. Refluxing 6
with in situ generated potassium ethylxanthate (∼150 mol
vs 6) in aqueous ethanol for a long period yielded
-mercaptoquinazolin-2-one (7) in 99.3% isolated yield.
%
2
Kilogram Processes of DL-028A and Impurity Profile
Determination (Scheme 3). Having solved the reagent issues
and recovery problems for the key intermediate 5, other
scaling-up issues were encountered during subsequent ma-
nipulations, such as environmental concerns, noxious byprod-
uct treatment, salt evaluation, and compound stability, among
others. In the sixth step, the N-alkylation reaction was
improved by the use of inexpensive piperazine hydrochloride
Notably, excess potassium ethylxanthate is essential to drive
the reaction to completion, otherwise, as shown by HPLC
monitor, the reaction stops within a few hours and with low
conversion. In the second step, the more nucleophilic
quinazolin-2-thione group of 7 was selectively methylated
with dimethyl sulfate in basic aqueous solution. After partial
neutralization of the resulting solution with a limited amount
of 3 N HCl (ca. 110 mol % vs 7) during the workup, the
unreacted 7 was left in the basic aqueous layer (pH ) 9-10)
in dissociated form, and the desired product 8 was selectively
isolated by filtration of the resulting slurry, to yield 8 in
1
0 (∼190 mol % vs 5) as the substitute for its free base
(Scheme 3), and the acid scavenger sodium bicarbonate was
added in two separate portions during refluxing to avoid the
elimination of HBr from 5 by excessive amounts of base.
After a long period of reflux, the cooled mixture was diluted
with water and then cooled to -11 °C with stirring for 3 h
to accelerate the precipitation of the desired product 11,
thereby eliminating the evaporation of the noxious solution.13
The isolated yield was 66.0%. In the seventh step, 11 was
hydrolyzed using excess aqueous sodium hydroxide in
methanol. The noxious methanethiol byproduct in alkaline
solution was destroyed by quenching with aqueous sodium
periodate before workup by evaporation of methanol and
subsequent manipulations. Sodium periodate (100 mol %)
82.5% isolated yield. The unreacted dimethyl sulfate was
quenched with ammonia water before disposal of wastewater.
In the third step, 8 was heated with phosphoryl chloride in
1,2-dichloroethane that contained a stoichiometric amount
of triethylamine to yield the chloride 9 as a lumpy solid,
which was left in the reactor and used directly in the next
step without filtration and purification. The fourth step was
readily completed in refluxing THF with allylamine (∼500-
7
00 mol % vs 9) after a short period (1 h, 60 °C versus 48
10,11
h, 120-130 °C, autoclave, original process ) as the
chloride 9 is more reactive to allylamine than is compound
3. Product 4 is a fine granular crystalline solid of high quality,
(13) The noxious smell is attributed to methanethiol generated by partial
and the isolation and washing processes were much easier
hydrolysis of 11.
274
•
Vol. 6, No. 3, 2002 / Organic Process Research & Development

Scheme 4. Decompostion of DL-028A
Table 1. Batch yields of kilogram-synthesis of DL-028A
batch starting material wt. final product wt.
overall yield
no. (anthranilamide) (kg) (DL-028A) (kg) (eight steps) (%)
K001
K002
K003
K004
1.00
1.00
2.00
2.00
0.668
0.596
1.038
1.050
19.6
17.5
15.2
15.4
although the content of 14 was usually insignificant (<0.1%).
During the course of the first kilogram run of DL-028A and,
as a result of laborious recrystallization from boiling 6/1 (v/
v) 2-propanol-water, only a 34.0% yield of DL-028A was
obtained, approximately 40.0% of the expected yield. The
labile property of the dihydroimidazole ring had also been
verified by an accelerated degradation experiment, in which
DL-028A was heated in boiling water (∼30 mL/g) for 16 h
and then analyzed by HPLC, indicating that the resulting
crude product consisted of unreacted DL-028 (15% by area)
and the ring-opened product 14 (85% by area), besides which
no other side products were detected. The structure of 14
was indirectly verified by NMR and mass spectra of an
acylation derivative 15 (Scheme 4). A limited amount (∼600
mol %) of concentrated hydrochloric acid was used as a
substitute for acetyl chloride, and the salt was formed at ca.
was initially added at 0 °C, and it digested the alkaline
solution for a long period (∼12 h) at that temperature.
However, a viscous and gel-like salt contaminated the
partially crystallized product 12 (DL-028). The salt could
not be removed from the crystalline solid by washing or
recrystallization or both. Therefore, the generated meth-
anethiol was quenched at room temperature using 100 mol
%
of aqueous sodium periodate, and the resulting mixture
was digested for 12 h at 35 °C. Thereafter, the salt was easily
removed by filtration, and no coprecipitated DL-028 was
1
4
observed. After evaporation of the pale odorous solvent,
the oily crude product was triturated with ethyl acetate to
give DL-028 free base as a fine powder. The isolated yield
was 81.0%. DL-028 free base is water-insoluble and required
conversion to an appropriate water-soluble salt before being
submitted to in vivo safety evaluation. Many forms of the
salt, such as citrate, hydrochloride, maleate, succinate, and
methanesulfonate have been prepared. Of these, only the
hydrochloride salt was sufficiently promising to produce
good solubility in water (∼0.12 g/mL), showing superior
crystallinity and high recovery during salt formation. Other
salts, such as citrate or methanesulfonate, were not easily
crystallized out from a variety of solvents and exhibited
inferior crystallinity. The maleate was sparingly soluble in
water, although the recovery of the salt from methanol was
excellent (∼95%). Acetyl chloride was initially used to
prepare methanolic hydrochloride and thus prepare kilogram
quantities of DL-028 hydrochloride. However, the volume
efficiency was low, and the crystallization rate was so high,
because of the limited solubility of DL-028 hydrochloride
in boiling methanol, that a highly uniform fine-powdered
crystalline solid could not be obtained. Moreover, laborious
multiple crystallization was needed. Surprisingly, during the
period of recrystallization from boiling 6/1 (v/v) 2-propanol-
water, the dihydroimidazole ring of DL-028 hydrochloride
was easily hydrolyzed to form a ring-opened byproduct 14
30 °C to prevent the hydrolysis of the dihydroimidazole ring
of DL-028A and thus avoid problems associated with thermal
instability of DL-028A with water, and volume efficiency.
Under such conditions, the operation volume was greatly
decreased, and the salt deposition rate was unexpectedly
moderated to deposit DL-028A as a fine, white powder. The
isolated yield was 83.4%. The overall yield of the eight-
step sequence was 15.2%, which yield was consistent to
within 5.0% deviation across four kilogram-scale runs (Table
1
). The levels of 14 in the stored kilogram products were
measured to be less than 0.1% and deemed acceptable for
1
6
use in preclinical study. Purity of a DL-028A reference
standard was determined through a series of analyses,
although it revealed a slightly hygroscopic property (see
Experimental Section).
Conclusions
The described reaction sequence constitutes a convenient
method for the kilogram-scale synthesis of DL-028 hydro-
chloride under normal conditions. The method also eliminates
some commonly encountered problems associated with large-
scale synthesis, such as hazardous reagents, dangerous
operations, multiple crystallizations, and others. Incidentally,
an in-process monitoring system and quality control for the
intermediates and final product were established. To extend
this work to larger scale (e.g., 100 kg), further studies, such
as rescale ability, scaleable workups, and waste-treatment
reevaluation, and so forth, are required. Results in this study
will hopefully form the basis of a manufacturing process.
(
Scheme 4), which was a viscous polar oil, whose structure
15
was confirmed by LC/MS spectrometry. Further experi-
mental results revealed that the hydrolysis was easily
executed at room temperature even by trace amounts of
moisture present in the solvent (e.g., MeOH or acetone),
Experimental Section
(14) No oxidized DL-028 byproducts were detected by HPLC and MS, due to
the stoichiometric amount of suspended sodium periodate used.
Reagents and solvents were obtained from commercial
1
3
sources and used as received. Proton and C magnetic
(
15) LC/MS: PE API 300; mobile phase: CH
3
CN:H
2
O ) 9:1 (v/v) + 0.1%
+
formic acid; flow rate: 5 µL/mL; direct inlet. Mass (m/z): 409.8 (M ),
+
), 217.8 (M+ - (2-methoxyphenyl)piperazinyl)), 188.8
3
(
92.3 (M - NH
3
(16) Owing to inferior operations in steps 7 and 8, the overall yield of the first
kilogram run (zero batch) was only 4.58%.
+
M - (2-methoxyphenyl)piperazinyl - H - C(O)).
Vol. 6, No. 3, 2002 / Organic Process Research & Development
•
275

resonance spectra were determined on a Bruker AC 500 MHz
NMR. The molar equivalent of hydrochloride in DL-028A
was determined using a Kyoto Electronics model AT-310
potentiometric automatic titrator. Melting points were gener-
ally measured in open capillary tubes using Buchi immersion
apparatus and are uncorrected. The melting point of DL-
added to the mixture at once. After 10 min of stirring the
clear solution turned turbid., after a further 8 h of stirring,
HPLC revealed that the area percent growth of 8 was less
than 1.0%. The solution was carefully treated with 3 N
hydrochloric acid (5.20 L, 15.6 mol) to pH ) 10.2 to
precipitate the desired product with the unreacted 7 that
remained left in the mother liquid. The precipitate was
collected and washed with water and then ethanol (3.0 L).
The obtained solid was dried in vacuo to give 2-methylth-
ioquinazolin-4(3H)-one (8) (2.31 kg, 82.5%) as an off-white
0
28A was measured using a Stream model DSC92 dif-
ferential scanning calorimeter. The water content of DL-
28A was determined using a Dynamic TGA 2959 thermal
0
gravimetric analyzer and a Kyoto Electronic MKC-210 Karl
Fischer moisture titrator. HPLC analyses were performed on
a Spectra-Physics HPLC system equipped with a Spectra-
Physics Focus Photodiode array detector and TSP Chron-
Quest integrator software. The HPLC method used for
analysis of intermediates (4, 5, 7-9, and 11) was performed
using an Inertsil ODS-2 column (250 × 4.6 mm) and a
acetonitrile/methanol/0.02 M phosphate buffer pH 4.5 (20/
18
powder, HPLC area 95.0%, mp 220-222 °C (lit. mp 213-
14 °C). An analytical sample was recrystallized from 95%
2
1
ethanol: mp 224-225 °C; H NMR (CDCl
3
) δ 11.3 (br s,
1
H, NH), 8.30 (dd, J ) 4.0, 1.4 Hz, 1H), 7.65 (d, J ) 7.9
+
Hz, 1H), 7.44 (m, 1H), 2.74 (s, 3H); MS (EI) m/z 192 (M );
Αnal. Calcd for C OS: C, 56.23; H, 4.19; N, 14.57.
Found C, 56.02; H, 4.08; N, 14.23.
-Chloro-2-methylthioquinazoline (9). In a 100-L glass
9 8 2
H N
4
30/50, v/v/v) isocratic mobile phase. The highly polar
reactor were successively placed 1,2-dichloroethane (23.0
L), 2-methylthioquinazolin-4(3H)-2-one (8) (2.31 kg, 12.03
mol), and triethylamine (1.70 L, 12.26 mol) before flushing
with nitrogen. After the solution was cooled to -2 °C,
phosphoryl chloride (4.6 L, 49.36 mol) was added at a rate
that kept the solution below 3.0 °C. After this period, the
mixture was refluxed for 2 h. After another 17 h of refluxing,
HPLC revealed that the area % growth of 9 was less than
intermediate 2 was determined using an alternative HPLC
method with the same column as that used in the above and
an acetonitrile/0.02 M phosphate buffer pH 4.5 (1/3, v/v)
isocratic mobile phase. The HPLC method used for analysis
of DL-028 (12) and DL-028A (13) was performed using a
COSMOSIL column (150 × 4.6 mm) and a methanol/0.01M
phosphate buffer pH 2.8 (30/70, v/v) isocratic mobile phase.
Preparative TLC (PTLC) was performed with precoated silica
gel plates, Merck 60 F254 (2.0 mm thickness). All the
procedures described below are consecutive manipulations
of a typical kilogram run (Batch K003, Table 1).
1.0%. The solution temperature was cooled to 3 °C, and then
2
6.8 L of water was added at a rate that kept the solution
below 10.0 °C. The organic layer was separated and the water
layer extracted twice with 1,2-dichloroethane (6 L × 2). The
combined organic layers were washed twice with 8 L of
water. The separated organic layer was added to the reactor
and evaporated. The solidified residue was dried in vacuo
to leave crude 4-chloro-2-methylthioquinazoline (9) as a
lumpy crystal, which was employed as the starting material
in the next step without purification: mp 104-105 °C (lit.12
2-Mercaptoquinazoin-2-one (7). In a 100-L glass reactor
were placed absolute ethanol (13.20 L) and potassium
hydroxide (85% purity, 1.45 kg, 22.04 mol), and the mixture
was heated to 25-30 °C to dissolve potassium hydroxide.
Carbon disulfide (1.68 kg, 22.04 mol) was added, and the
mixture was stirred at 25-30 °C for 0.5 h to produce a
yellow xanthate salt suspension. Anthranilamide 6 (2.00 kg,
1
07-108 °C). An analytical sample was recrystallized from
14.68 mol), 95% ethanol (15.20 L), and water (2.24 L) were
isopropyl ether to give 9 as a white powder: mp 106-107
then added to the mixture in succession. The progress of
the reaction was monitored by HPLC every 1-2 h. After
1
°
3
C; H NMR (CDCl ) δ 8.16 (d, J ) 8.2 Hz, 1H), 7.87 (d,
J ) 3.8 Hz, 2H), 7.56 (m, 1H), 2.68 (s, 3H); MS (EI) m/z
21 h of reflux, HPLC revealed that the area % growth of 7
+
2
11 (M + 1); Αnal. Calcd for C
9
H
7
ClN
2
S: C, 51.31; H,
was less than 0.3%. Refluxing was discontinued, and the
solution was cooled to 25 °C before being diluted with 20 L
of water and then cooled to 3 °C. The resulting solution was
carefully treated with 5 L of glacial acetic acid and stirred
for 16 h at that temperature. The precipitate was filtered,
washed with ice water, and dried in vacuo to give 2-mer-
captoquinazoin-2-one (7) (2.60 kg, 99.3%) as an off-white
powder, HPLC area 99.7%. An analytical sample was
3
.35; N, 13.30. Found C, 51.13; H, 3.34; N, 13.25.
4-Allylamino-2-methylthioquinazoline (4). To the 100-L
glass reactor that contained crude 9 (ca. 2.0 kg), as described
in the above procedure, was added THF (24.2 L) before
stirring to produce an homogeneous solution. To the mixture
was added allylamine (6.70 L, 89.0 mol) before refluxing
for 1 h. HPLC showed less than 1.0% unreacted 9. After
this period, the re-cooled mixture was evaporated in vacuo
to remove THF. The residue was cooled to 0-5 °C, ice water
recrystallized from 95% ethanol to give a white powder: mp
1
>
295 °C; H NMR (d
6
-DMSO) δ 7.94 (d, J ) 6.9 Hz, 1H),
(21.40 L) was added, and vigorous stirring was continued
7
.73 (m, 1H), 7.37 (d, J ) 8.2 Hz, 1H), 7.32 (m, 1H).; MS
for 1 h at 5 °C. The precipitate was collected on a centrifuge
filter, washed with a small amount of ice-cooled acetonitrile
+
(EI) m/z 178 (M ); Αnal. Calcd for C
8 6 2
H N OS: C, 59.04;
H, 3.72; N, 17.23. Found C, 58.87; H, 3.56; N, 16.93.
(3.00 L), and then dried in vacuo to give 4-allylamino-2-
2-Methylthioquinazolin-4(3H)-one (8). In a 100-L glass
methylthioquinazoline (4) (1.41 kg, 50.9%, two steps) as a
reactor were successively placed crude 7 (2.60 kg, 14.6 mol),
water (38.0 L) and sodium hydroxide (93% purity, 1.00 kg,
10,11
white powder: HPLC area 100.0%, mp 130-131 °C (lit.
(
17) DL-028A DeVelopment Abstract, Feb 06, 2001 (DCB unpublished file).
23.25 mol). Heating to 25-30 °C dissolved the sodium
(18) Hu, M.-K.; Liu, K.-C.; Hsu, L.-Y.; Shih, B.-J. Chin. Pharm. J. 1991, 43,
hydroxide. Dimethyl sulfate (2.04 kg, 16.18 mol) was then
83.
276
•
Vol. 6, No. 3, 2002 / Organic Process Research & Development

1
1
33-135 °C). An analytical sample was recrystallized from
/1 (v/v) ethyl acetate-hexane: mp 132-134 °C; H NMR
before stirring for 30 min. To the mixture, a solution of
sodium hydroxide (assay: 93%) (4.26 kg, 99.00 mol) in
water (34.3 L) was added, and after 20 h of refluxing, HPLC
revealed that the area % growth of 12 was less than 0.5%.
The solution was cooled to 35 °C, and a solution of sodium
periodate (0.71 kg, 3.31 mol) in water (6.3 L) was added
and stirred at that temperature for 17 h to quench the
methanethiol byproduct. The resulting mixture was filtered
to remove viscous suspensions and washed with methanol,
and the filtrate was concentrated under reduced pressure to
remove methanol. The residue was diluted with dichlo-
romethane (10.0 L) with stirring; then the organic layer was
collected and successively washed with saturated brine (18.0
L) and finally with water. The separated organic layer was
dried and evaporated, and the residual oil was diluted with
ethyl acetate (9.4 L) and stirred vigorously at 15 °C for 16
h to allow the oily suspension to crystallize and thus produce
a white powder. The precipitate was collected by filtration
and dried in vacuo to yield 12 (i.e., DL-028) as a white
powder (1.04 kg, 81.0% yield): HPLC area 98.7%, mp 210-
1
3
(CDCl ) δ 7.66-7.72 (m, 3H), 7.33 (m, 1H), 6.04 (m, 1H),
5.92 (br s, 1H), 5.33 (d, J ) 17.1 Hz, 1H), 5.23 (d, J ) 10.2
+
Hz, 1H), 4.32 (m, 2H), 2.65 (s, 3H); MS (EI) m/z 231 (M ),
2
05; Αnal. Calcd for C12
Found C, 62.17; H, 5.37; N, 18.05.
-Bromomethyl-5-methylthio-2,3-dihydroimidazo[1,2-
13 3
H N S: C, 62.31; H, 5.66; N, 18.17.
3
C]quinazoline (5). In a 100-L glass reactor were placed
acetonitrile (16.3 L) and crude 4-allylamino-2-methylthio-
quinazoline (4) (1.41 kg, 6.10 mol). The mixture was stirred
for 10 min. N-bromosuccinimide (1.29 kg, 7.24 mol) was
added, and the reaction was stirred at room temperature. After
1
h of stirring, HPLC showed less than 1.0% unreacted 4.
The solution was cooled to 3 °C and stirred for 1 h at that
temperature and then filtered; the precipitate was washed
with a small amount of ice-cooled acetonitrile (3.0 L) and
dried in vacuo to yield 3-bromomethyl-5-methylthio-2,3-
dihydroimidazo[1,2-C]quinazoline (5) (1.55 kg, 82.0%) as
11,12
a white powder: HPLC area 99.2%, mp 162-163 °C (lit.
62-163 °C). An analytical sample was recrystallized from
1
0,11
1
1
211 °C (lit.
211-212 °C). H NMR (d -DMSO) δ 7.79
6
1
acetone: mp 161-162 °C; H NMR (CDCl
.8 Hz, 1H), 7.55 (m, 1H), 7.43 (d, J ) 8.0 Hz, 1H), 7.27
m, 1H), 4.64 (m, 1H), 4.26 (m, 1H), 4.09 (dd, J ) 7.5, 3.6
3
) δ 7.98 (d, J )
(d, J ) 7.5 Hz, 1H), 7.49 (m, 1H), 7.08 (m, 2H), 6.94 (m,
2H), 6.87 (m, 2H), 4.48 (m, 1H), 4.04 (m, 1H), 3.90 (dd, J
) 7.5, 5.0 Hz, 1H), 3.77 (s, 3H), 3.37 (m, 4H), 2.95 (m,
7
(
+
Hz, 1H), 3.75 (m, 1H), 3.62 (m, 1H), 2.67 (s, 3H); MS (EI)
m/z 310 (M ), 230, 174; Αnal. Calcd for C12
4H), 2.83 (m, 1H), 2.67 (m, 2H); MS (EI) m/z 391 (M ),
+
12 3
H N SBr: C,
25 5 2
376; Αnal. Calcd for C22H N O : C, 67.50; H, 6.44; N,
4
6.46; H, 3.90; N, 13.55. Found C, 46.22; H, 3.68; N, 13.26.
-{4-[1-(2-Methoxyphenyl)piperazinyl]}methyl-5-meth-
ylthio-2,3-dihydroimidazo[1,2-C]quinazoline (11). In a
00-L glass reactor were successively placed 1-(2-methoxy-
17.89. Found C, 67.45; H, 6.37; N, 17.77.
3
3-{4-[1-(2-Methoxymethyl)piperazinyl]}methyl-2,3-di-
hydroimidazo[1,2-C]quinazolin-5(6H)-one Dihydrochlo-
ride (13) (i.e., DL-028A). Into a 100-L glass reactor were
placed DL-028 (1.04 kg, 2.66 mol) and methanol (55.0 L),
and the mixture was heated with stirring at 45 °C for 30
min to produce a solution. The solution was cooled to 30-
32 °C, and concentrated hydrochloric acid (37%, 1.47 L,
14.9 mol) was added at a rate that kept the solution at 30-
35 °C. After this period, the solution was cooled to room
temperature over a 10-h period and stirred at a moderate
rate for another 68 h, after which time, a large amount of
white powder was deposited. The precipitate was filtered,
washed with ice-cooled methanol, and dried in vacuo to yield
1
phenyl)piperazine hydrochloride 10 (2.16 kg, 9.44 mol),
sodium bicarbonate (1.19 kg, 14.20 mol), and acetonitrile
(23.0 L) before stirring for 30 min. Then a solution 5 (1.55
kg, 5.00 mol) in acetonitrile (11.3 L) was added to the
mixture. After 20 h of refluxing, the solution temperature
was allowed to decline to 35 °C, and a second charge of
sodium bicarbonate (1.26 kg, 15.04 mol) was added, with
continued refluxing for 14 h. Then the reaction was
monitored every 1 h. After another 5 h of refluxing, HPLC
showed less than 1.0% unreacted 10. The solution was cooled
to 25 °C and the mixture diluted with water (32.0 L) and
stirred vigorously. After the solution was cooled to -11 °C
and stirred for 3 h, the precipitate was collected by filtration
and dried in vacuo to yield 11 (1.39 kg, 66.0% yield) as a
13 as a white powder (1.04 kg, 83.4% yield): HPLC area
1
100.0%, mp 254-255 °C dec; H NMR (D
2
O) δ 8.00 (d, J
) 8.0 Hz, 1H), 7.92 (m, 1H), 7.46 (m, 1H), 7.38 (d, J ) 8.5
Hz, 1H), 7.26 (m, 1H), 7.23 (m, 1H), 7.11 (d, J ) 8.0 Hz,
1H), 7.04 (m, 1H), 5.48 (br s, 1H), 4.56 (t, J ) 11.9 Hz,
1H), 4.11 (dd, J ) 6.3, 6.3 Hz, 1H), 3.96 (m, 1H), 3.88 (s,
3H), 3.79 (br s, 2H), 3.68 (m, 1H), 3.60 (m, 8H).
10,11
white powder: HPLC area 100.0%, mp 170-172 °C (lit.
1
2
74-175 °C). An analytical sample was recrystallized from
1
-butanone: mp 173-175 °C; H NMR (d -DMSO) δ 7.84
6
(
d, J ) 7.8 Hz, 1H), 7.57 (m, 1H), 7.35 (d, J ) 8.1 Hz, 1H),
.28 (m, 1H), 6.94 (m, 2H), 6.88 (m, 2H), 4.55 (m, 1H),
.07 (m, 1H), 3.93 (dd, J ) 7.4, 3.2 Hz, 2H), 3.77 (s, 3H),
Preparation and Purity Determination of DL-028A
Reference Standard. Preparation. A set of 100 sample vials
was prepared by dividing an appropriate amount of purified
DL-028A, prepared by recrystallization of the product from
the kilogram run from boiling 6/1 (v/v) 2-propanol/water;
50 mg of the purified DL-028A was transferred to each vial
and placed into an oven in vacuo at 70 °C for 16 h.
Analytical Experiment and Purity Definition. Problems
were experienced during measurement with the EA (elemen-
tal analyzer) because DL-028A absorbed moisture during the
analytical process. Accordingly, these samples were treated
7
4
2
.97 (m, 4H), 2.76 (m, 3H), 2.61 (s, 3H), 2.55 (m, 3H); MS
+
(EI) m/z 422 (M + 1), 374; Αnal. Calcd for C23
H N
27 5
SO:
C, 65.53; H, 6.46; N, 16.61. Found C, 65.46; H, 6.38; N,
6.46.
-{4-[1-(2-Methoxyphenyl)piperazinyl]}methyl-2,3-di-
hydroimidazo[1,2-C]quinazolin-5(6H)-one (12) (i.e., DL-
28). In a 100-L glass reactor were successively placed
compound 11 (1.39 kg, 3.30 mol) and methanol (27.8 L)
1
3
0
Vol. 6, No. 3, 2002 / Organic Process Research & Development
•
277

by the following procedure before analysis using the EA and
TGA (thermal gravimetric analyzer). Vials from the above
samples were placed into a desiccator at humidity RH 72%
for 20 h, and vials were randomly selected for measurement
with TGA and EA. The reported TGA and EA data are the
mean measurements for the three random samples. The
residual solvent (2-propanol) was not detected by standard
gas chromatography analysis.19 The acid-base molar ratio
of DL-028A was determined to be 2.00 by a potentiometric
automatic titrator. Thus, the water content of the reference
standard can be given by the following formula:
Hydrolysis of DL-028A: An Accelerated Preparation
of the Ring-Opening Product 14 and Acylation Derivative
15 by Thermal Hydrolysis of DL-028A (13). A mixture of
DL-028A (13) (1.0 g, 2.15 mmol) and water (30 mL) was
heated under reflux for 16 h and then evaporated in vacuo
to leave an oily residue. An aliquot was neutralized and
subjected to LC/MS analysis. The oily residue was dis-
solved in 1,2-dichloroethane (30 mL), cooled to 0 °C, and
successively treated with triethylamine (0.84 g, 8.31 mmol)
and acetic anhydride (0.32 g, 3.13 mmol). After stirring for
15 h at room temperature, the resulting solution was washed
successively with water, 5% NaOH, and water. The separated
1
5
1
8.0152x
4
organic layer was dried (MgSO ) and evaporated to leave
an oily residue (0.6 g). An aliquot of the crude product was
purified on repeated PTLC using 2% methanol in ethyl
(
w/w %) of H O )
2
4
64.3936 + (18.0152x)
in which DL-028A with H
xH O, (w/w %) of H O is weight percentage of water in DL-
28‚2HCl‚xH O. The purity of DL-028A (the reference
2
O is formulated as DL-028‚2HCl‚
acetate to isolate 15 from recovered DL-028 as a viscous
2
2
1
oil (lower R
f
portion). H NMR (CDCl
3
) δ 9.25 (br s, 1H,
0
2
NHC(O)N, varies with concentration), 8.10 (br s, 1H), 7.56
standard) is shown by this formula:
(m, 1H), 7.20 (m, 1H), 7.05 (d, J ) 8.0 Hz, 1H), 6.99 (m,
1
1
H), 6.83-6.96 (m, 3H), 6.51 (br s, NHAc), 5.25, 5.37 (m,
C - C
i
0
20
H, C(O)NCHC(O)), 3.70, 3.85 (each m, 2H, CH
), 3.04 (m, 6H, CH (piperazinyl) +
piperazinyl H), 2.77 (m, 4H, piperazinyl H), 1.93 (br s, 3H,
2
NHAc),
Purity of DL-028A ) 1 -
(
)
Ci
3.84 (s, 3H, OCH
3
2
in which C
xH O determined by an EA, C
of DL-028‚2HCl‚xH O.
Analytical Data for DL-028‚2HCl‚xH
content (i.e., (w/w %) of H O) was determined to be 6.7%
by TGA or a Karl Fischer titrator; therefore
0
is an observed value (C%) of DL-028‚2HCl‚
13
3 3
NHC(O)CH ); C NMR (CDCl ) δ 170.8, 163.6, 152.3,
2
i
is the theoretical value (C%)
141.1, 139.3, 135.2, 133.5, 128.7, 128.0, 126.5, 123.2, 121.1,
118.3, 115.3, 111.3, 76.9, 60.1, 59.4, 59.3, 58.6, 54.5, 53.8,
50.8, 49.6. LC/MS: PE API 300; mobile phase: MeOH;
2
2
O. The water
2
flow rate: 25 µL/mL; direct infusion. Mass (m/z): 452.0
+
+
(M +1), 392.0 (M - NHAc - H).
Acknowledgment
1
8.0152x
0
.067 )
,
x ) 1.861
We thank the Ministry of Economic Affairs, R. O. C. for
financially supporting this research under Project No. MOEA
88-2-3 and MOEA 89-2-3. Professor Ji-Wang Chern is
appreciated for his valuable discussions and encouragement.
We also thank Miss Mei-Huey Lee for performing NMR
and Mass spectra.
4
64.3936 + (18.0152x)
Elemental Analysis: Calcd for DL-028‚2HCl‚1.861-
O: C, 53.07; H, 6.22; N, 14.07. Found: C, 52.83; H, 5.99;
N, 13.89. Therefore C and C were determined as 53.07 and
0 i
H
2
5
2.83, respectively.
Purity Determination for DL-028A Reference Standard.
According to the above purity definition, the purity of the
DL-028A reference standard ) 1 - ((53.07-52.83)/53.07)
Received for review September 24, 2001.
OP0100807
)
99.55%.
(
20) The unusual split low-field shifts of NCH are attributed to highly electron-
withdrawing and enolisable neighboring groups, a related NCH (δ 5.08)
1
0
(19) USP organic volatile impurity (467 method V).
appeared in a 3,4-dihydro-4-iminoquinazolinone derivative.
278
•
Vol. 6, No. 3, 2002 / Organic Process Research & Development