Chemical development of ZD9331: Synthesis of a bromomethylquinazolinone avoiding a nonselective radical bromination

Article abstract of DOI:10.1021/op060049a

An efficient regiospecific synthesis of ZD9331 Pivaloyloxymethyl (POM) Bromide (4) has been accomplished via ZD9331 Quinacetate HCl (15) avoiding a nonselective bromination. The original route used a radical bromination on a substrate with three methyl groups, which generated a range of bromomethyl derived compounds that carried through to the final active pharmaceutical ingredient (API). A strategy, based on the Zinin reaction, was developed to synthesize the required bromomethyl compound in a regioselective manner. This approach was successfully scaled to manufacture a tonne of material.

Full text of DOI:10.1021/op060049a

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Organic Process Research & Development 2006, 10, 553−555
Technical Note
Chemical Development of ZD9331: Synthesis of a Bromomethylquinazolinone
Avoiding a Nonselective Radical Bromination
Dagmar Bentley, Andrew A. Godfrey,* and Kenneth E. H. Warren
Process Research and DeVelopment Department, AstraZeneca, Silk Road Business Park,
Charter Way, Macclesfield, Cheshire SK10 2NA, UK
Abstract:
An efficient regiospecific synthesis of ZD9331 Pivaloyloxymethyl
(POM) Bromide (4) has been accomplished via ZD9331 Quinace-
tate HCl (15) avoiding a nonselective bromination. The original
route used a radical bromination on a substrate with three
methyl groups, which generated a range of bromomethyl
derived compounds that carried through to the final active
pharmaceutical ingredient (API). A strategy, based on the Zinin
reaction, was developed to synthesize the required bromomethyl
compound in a regioselective manner. This approach was
successfully scaled to manufacture a tonne of material.
Introduction
ZD9331 (1) (Figure 1) is a nonpolyglutamable thymidalate
synthase (TS) inhibitor, for the potential treatment of gastric
cancer and solid tumors. Raltitrexed (2) (Figure 1) had
already been launched as a TS inhibitor for the treatment of
Figure 1.
colorectal cancer; ZD9331 has a range of advantages over
Raltitrexed, centered mainly around increased potency, by
virtue of a more effective transport mechanism to solid
tumors. ZD9331 originated from a collaboration between
Zeneca, the Institute of Cancer Research (Sutton, UK), and
BTG International Limited (BTG).1 Details of ZD9331’s
pharmacological profile are cited in a previous paper.2
A particular problem with the original medicinal chemistry
route3 (Scheme 1) was a radical bromination of ZD9331
POM Quinazolinone (3), predominantly giving the bromo-
methyl intermediate 4; but as there are three positions capable
of undergoing a free radical bromination, a mixture of
regioisomeric compounds was isolated after the subsequent
alkylation of ZD9331 Propargylaniline (8) (Scheme 2).
ZD9331 Pyrroloquinazolinone (17), derived from the dibro-
minated intermediate 6, was particularly problematic as it is
highly crystalline; hence its concentration increased through
subsequent processing. As the final active pharmaceutical
ingredient (API) is amorphous, control of 17 in the isolated
API was extremely difficult in early scale-up. No effective
method to separate it from ZD9331 POM Quinazolinone
ester (7) could be devised without recourse to chromatog-
raphy. The regioisomer derived from intermediate 5 was lost
to the mother liquors upon isolation of 7. The design of an
alternative synthetic strategy was constrained by the fact that
ZD9331 had already been subject to extensive toxicological
screening, so any new impurities would have to be qualified
by further toxicity testing; any new approach would have to
meet this demanding constraint.
Results and Discussion
Retrosynthetic analysis of the ZD9331 POM Bromide (4)
(Scheme 3) identified that the methyl group, requiring
activation as an electrophile, was para to an amino group,
potentially accessed by reduction of a nitro group. Such an
arrangement is set up to use a known feature of the Zinin
reaction, first published in 1842, where a nitro group can be
reduced, and concomitantly a para or ortho methyl group is
oxidized to an aldehyde in one step using sodium polysulfide
in aqueous sodium hydroxide.4,5 Compound 9 was prepared
* To
andy.godfrey@astrazeneca.com.
(1) Godfrey, A. A. WO 2005012260, 2004, p 29.
whom
correspondence
should
be
addressed.
Email:
(2) Moseley, J. D.; Bansal, P.; Bowden, S. A.; Couch, A. E. M.; Hubacek, I.;
Weingartner, G. Org. Process Res. DeV. 2006, 10 (1), 153-158.
(3) Marsham, P. R.; Wardleworth, J. M.; Boyle, F. T.; Hennequin, L. F.;
Kimbell, R.; Brown, M.; Jackman, A. L. J. Med. Chem. 1999, 42 (19),
3809-3820.
(4) Porter, H. K. Org. React. 1973, 20, 455-481.
(5) Zinin, N. J. Prakt. Chem. 1842, 27, 149.
10.1021/op060049a CCC: $33.50 © 2006 American Chemical Society
Published on Web 04/15/2006
Vol. 10, No. 3, 2006 / Organic Process Research & Development
•
553

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Scheme 1
Scheme 2
9 through to the desired ZD9331 POM Bromide (4). The
synthesis was designed by adapting the previous strategy of
bromination and then functional group interconversion to a
nitrile, followed by ring closure under acidic conditions to
give the quinazolinone skeleton8 (Scheme 4). A review of
the synthesis of quinazolinones has recently been published
by Patrick J. Guiry et al.9 Initially it was hoped that the
aldehyde group would be amenable to this strategy, ulti-
mately allowing coupling via a reductive amination, rather
than an alkylation of 8. Bromination of 9 gave a 60:40
mixture of the 5- and 6-bromo compounds, so this approach
was abandoned. A practical strategy (Scheme 4) was
designed by reduction of 9 to 10 with sodium borohydride,
as described by Youichi Shiokawa et al.10 followed by
acetylation of both the alcohol and amino groups to give
intermediate 11. Bromination of 11 gave an 85:15 mixture
of the 3- and 5-bromo compounds; fortunately, 13 was lost
to the mother liquors on workup, because otherwise this
strategy would not have been any more efficient than the
original free radical bromination approach. 1,3-Dibromo-5,5-
dimethylhydantoin was the preferred reagent over N-bro-
mosuccinimide as it has two available bromines. Acetylation
and bromination were telescoped into one stage to improve
the overall efficiency. Cyanation of 12 with copper cyanide
in dimethylformamide (DMF) gave a relatively low strength
for compound 14 (83% w/w), which was contaminated with
unreacted 12 and residual solvent. However 12 could be
removed in subsequent crystallisations. Ring closure of 14
with hydrogen chloride in 2-propanol gave 15 as a single
regioisomer, although care had to be taken to prevent ingress
of moisture as this led to high levels of the related
hydroxymethyl compound. As described elsewhere2 after
protection with a POM group, the acetoxy compound was
reacted with hydrogen bromide in acetic acid to yield 4. As
expected, 4 reacted cleanly with 8 to give the key intermedi-
ate 7.
Scheme 3
using the procedure well exemplified in the literature;6 an
insight into the mechanism of the reaction is given in the
paper by Yoshiro Ogata et al.7 The challenge was to elaborate
Scheme 4
554
•
Vol. 10, No. 3, 2006 / Organic Process Research & Development

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Conclusion
4-(acetylamino)-3-bromo-2-methylphenyl acetate (13) was
lost to the aqueous acetonitrile wash to yield the product
ZD9331 Bromide (12, 56 g 62%) with HPLC purity typically
98% w/w.
Thus exploiting 19th century chemistry overcame an
inherent weakness in the first scaled synthesis, namely an
impurity that increases in concentration through subsequent
processing, always a difficult challenge, but particularly in
this case where the final API is amorphous.
1H NMR δ (DMSO-d6): 2.1 (s, 3H), 2.2 (s, 3H), 2.3 (s,
3H), 5.0 (s, 2H), 7.6 (s, 1H) 7.4 (s, 1H), 9.5 (s, 1H).
ZD9331 Nitrile (14). ZD9331 Bromide (12, 50 g, 0.167
mol), copper(I) cyanide (14.2 g, 0.159 mol), and dimethyl-
formamide (100 mL) were heated at 90 °C under an
atmosphere of nitrogen. After 6 h the mixture was cooled to
60 °C and treated portionwise with zinc powder (13.1 g, 0.2
mol), and then the slurry was reheated to 90 °C, screened
through Celite, cooled to 50 °C, and diluted with water (400
mL). On cooling to 20 °C the product was isolated by
filtration, washed with water, and dried to constant weight
at 50 °C in vacuo to yield the product ZD9331 Nitrile (14,
41.4 g 84%) with HPLC purity typically 83% w/w.
1H NMR δ (DMSO-d6): 2.1 (s, 3H), 2.25 (s, 3H), 2.5 (s,
3H), 5.3 (s, 2H), 7.6 (s, 1H), 7.9 (s, 1H), 10.3 (s, 1H).
ZD9331 Quinacetate HCl (15). Hydrogen chloride gas
(0.12 kg, 3.29 mol) was added over 60 min to a slurry of
ZD9331 Nitrile (14, 0.67 kg, 2.7 mol) in propan-2-ol (6.7
L). On cooling to 30 °C ZD9331 Quinacetate HCl (15)
crystallised out of solution. The product was isolated by
filtration, washed with propan-2-ol, and dried to constant
weight at 50 °C in vacuo to yield the product ZD9331
Quinacetate HCl (15, 0.662 kg 87%) with HPLC purity
typically 94% w/w.
Experimental Section
Reagents were purchased from standard suppliers.
NMR spectra were run at 270 MHz in d6-DMSO solution
and are reported in parts per million downfield from internal
TMS.
HPLC analyses were conducted using a HiChrome RPB
column 25 cm × 0.46 cm, solvent system acetonitrile/water/
TFA 1200/800/1 (v/v/v), flow rate 1 mL/min, and detection
at 235λ.
ZD9331 Bromide (12). Triethylamine (63 mL, 0.45 mole)
was added in one portion to a slurry of ZD9331 Alcohol
(10, 54 g, 0.3 mol equiv), in ethyl acetate (540 mL) at
ambient temperature. The slurry was heated to 50 °C, acetyl
chloride (30 mL, 0.42 mol equiv) was added over 2 h, and
after a further 30 min the mixture was cooled to 20 °C. The
slurry was extracted sequentially with water (2 × 270 mL)
and saturated brine (270 mL). The ethyl acetate extract was
solvent swapped into acetonitrile by distillation. The aceto-
nitrile solution of 4-(acetylamino)-2-methylphenyl acetate
(11) was treated with a solution of 1,3-dibromo-5,5-dimeth-
ylhydantoin (48.6 g, 0.17 mol equiv) in acetonitrile (380 mL)
at 50 °C, and after 60 min the reaction mixture was cooled
to 20 °C and poured into water (1350 mL). ZD9331 Bromide
(12) was isolated by filtration, washed with water, and dried
to constant weight at 50 °C in vacuo. The regioisomer
1H NMR δ (DMSO-d6): 2.1 (s, 3H), 2.4 (s, 3H), 2.7 (s,
3H), 5.2 (s, 2H), 7.7 (s, 1H), 8.1 (s, 1H).
Acknowledgment
We are grateful to BTG International Ltd for permission
to disclose this research.
(6) Burgess, D. A.; Rae, I. D. Aust. J. Chem. 1977, 30, 927-931.
(7) Ogata, Y.; Kawasaki, A.; Sawaki, Y.; Nakagawa, Y. Bull. Chem. Soc. Jpn.
1979, 52 (8), 2399-2401.
(8) Showell, G. A. Synth. Commun. 1980, 10 (3), 241-243.
(9) Connolly, D. J.; Cusack, D.; O’Sullivan, T. P.; Guiry, P. J. Tetrahedron
2005, 61 (43), 10153-10202.
Received for review February 28, 2006.
OP060049A
(10) Shiokawa, Y.; Nagano, M.; Itani, H. EP 268989, 1987, p 59.
Vol. 10, No. 3, 2006 / Organic Process Research & Development
•
555