Process development and multikilogram syntheses of XL228 utilizing a regioselective isoxazole formation and a selective SNAr reaction to a pyrimidine core

Article abstract of DOI:10.1021/op400137m

Route scouting, process development, and multikilogram syntheses of an IGF-1R/Src/Bcr-Abl inihibitor are reported. Key aspects of the developed route are a regioselective [3 + 2] isoxazole formation on a pyrimidine core and a selective S_NAr addition of an aryl amine to a symmetrical dichloro substituted pyrimidine. The route contains six synthetic steps and was demonstrated twice on scale, delivering 4.6 and 11.2 kg (25% and 16% overall yield), for Phase I clinical studies.

Full text of DOI:10.1021/op400137m

Communication
pubs.acs.org/OPRD
Process Development and Multikilogram Syntheses of XL228
Utilizing a Regioselective Isoxazole Formation and a Selective SNAr
Reaction to a Pyrimidine Core
Nathan R. Guz,*,† Helena Leuser,§,⊥ and Erick Goldman‡,†
†Chemical Development, Exelixis, Inc., South San Francisco, California 94080, United States
§Chemical Development, CARBOGEN-AMCIS, CH-5001 Aarau, Switzerland
S
* Supporting Information
substrate in the SNAr additions of amines, phenols, thiols, and
other nucleophiles.6 Using established precedent,7 the SNAr
ABSTRACT: Route scouting, process development, and
multikilogram syntheses of an IGF-1R/Src/Bcr-Abl
inihibitor are reported. Key aspects of the developed
route are a regioselective [3 + 2] isoxazole formation on a
pyrimidine core and a selective SNAr addition of an aryl
amine to a symmetrical dichloro substituted pyrimidine.
The route contains six synthetic steps and was
demonstrated twice on scale, delivering 4.6 and 11.2 kg
(25% and 16% overall yield), for Phase I clinical studies.
addition between 3 and 2 was quickly optimized (8 vol of IPA,
1.5 equiv of TEA, 20 °C, 3 h then 14 vol of H2O at 20 °C) on
small scale and demonstrated on a 100-g scale to yield 4 in 89%
yield and 98.5% purity with <0.1% of 4a and bis-addition
products present. This SNAr reaction showed similar perform-
ance using 1-butanol, N,N-dimethylacetamide (DMA), or
CH2Cl2 as the solvent and N,N-diisopropylethylamine
(DIPEA) as the base.
Satisfied with a repeatable and scalable procedure to
synthesize 4, a broad array of solvents (DMA, IPA, 1-butanol,
CH2Cl2, heptane/water, toluene, THF, 2-MeTHF), temper-
atures (45 to 80 °C), and bases (TEA, DIPEA, DBU, Na2CO3,
K2CO3) were investigated to promote the selective addition of
5 to 4. Acid-mediated SNAr additions (HCl, ZnBr2)8 of 5 to 4
were also examined. The multitude of reaction conditions
investigated consistently returned 6 with 14−25% of 6a.
Recrystallizations and slurries from MtBE/MeOH and iPrOAc/
MeOH were able to partially remove 6a from 6, but 6−14% of
the undesired 6a remained, and isolated yields of 6 were no
more than 40%. The purest sample of 6 (with 5.8% 6a)
submitted to the subsequent SNAr reaction with 7 produced 1,
along with the corresponding isomer of 1 (7 added to the 2-
position of 6a). Attempts to purify 1 using various iPrOAc/
MtBE, iPrOAc/MtBE/MeOH, and iPrOAc/MtBE/EtOH
recrystallization systems failed to provide 1 in greater than
97.5% purity.
The inefficient use of the expensive subunit 5 (∼$4000/kg,
6−8 week lead time) in the modified medicinal chemistry route
(Scheme 1) coupled with the difficulties associated with
purifying 1 to >99% warranted a new approach to a
kilogram-scale process.
Second-Generation Approach to 1. To overcome the
poor selectivity encountered with the SNAr addition of 5 to 4, a
new method to attach the substituted isoxazole subunit to the
pyrimidine ring was necessary. During the process development
of the modified medicinal chemistry route (Scheme 1), [3 + 2]
cycloaddition reactions were explored in-house as a method to
synthesize 5. Scheme 2 describes one method that proved to be
an efficient and cost-effective approach to 5.9
INTRODUCTION
■
Pyrimidine 1 is a potent, reversible, and ATP competitive small-
molecule inhibitor developed in 2004 at Exelixis, Inc. This
molecule targets the IGF-1R, Src, and Bcr-Abl pathways. IGF-
1R and Src both play important roles in cell proliferation and
survival in colon, breast, prostate, ovarian, lung, and
hepatocellular cancers.1,2 The Bcr-Abl gene is responsible for
most patients afflicted with the hematological malignancies
CML (chronic myelogenous leukemia), ALL (acute lympho-
blastic leukemia), and AML (acute myelogenous leukemia).3
XL228 also targets the T315I mutant form of Bcr-Abl, a mutant
that shows resistance to conventional Bcr-Abl therapies.4
Discovery Synthesis. The medicinal chemistry approach
to synthesize 1 (Scheme 1) utilized 2 as the core building block
for three separate nucleophilic aromatic substitution (SNAr)
reactions. This short synthesis was capable of producing 35 g of
1 for toxicological range-finding studies, but chromatography or
multiple hot slurry procedures were necessary to purify the
intermediates and 1 to high-performance liquid chromatog-
raphy (HPLC) purities of >95%. These purification techniques
would not be amenable for manufacturing the approximately 4
kg of 1 necessary to initiate IND enabling toxicology and Phase
I clinical studies. However, the brevity of the medicinal
chemistry synthesis was quite attractive, and initial route
optimization for a multikilogram synthesis of 1 focused on this
approach.5
RESULTS AND DISCUSSION
■
Initial Route Optimization. In order for the route
presented in Scheme 1 to be successful in providing kilogram
quantities of 1, the SNAr reactions would need to be selective
and the resulting products amenable to straightforward
crystallizations. The pyrimidine 2 is well investigated as a
Received: May 29, 2013
© XXXX American Chemical Society
A
dx.doi.org/10.1021/op400137m | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
a
Scheme 1. Medicinal chemistry route
a
Reagents and conditions: (a) 1-BuOH, TEA, 80 °C, 2 h, 92%, 11.8:1 4/4a; (b) 1-BuOH, TEA, 120 °C, 16 h, 39%, 3.1:1 6/6a; (c) 110 °C, 3 h, 69%.
a
Scheme 2. In-house method to synthesize 5
a
Reagents and conditions: (a) 50% NH2OH in H2O, 20 °C, 8 h, 90%; (b) N-chlorosuccinimide, DMF, 20 °C, 12 h; (c) CH2Cl2, TEA, 20 °C, 2 h;
(d) 4 N HCl/IPA, 0 °C, 12 h, 74% from 8.
a
Scheme 3. Isoxazole ring formation
a
Reagents and conditions: (a) Ac2O, DMA, 130 °C, 5 h, 67%; (b) propargyl bromide (80% in PhMe), K2CO3, DMA, 40 °C, 16 h, 79%; (c) 9, DMA,
40 °C, 10 h, 97% conversion by LC.
The syntheses of 8 and 9 were straightforward, and the
addition of Boc-propargylamine to 9 produced 5 as a single
regioisomer (both 5 and reaction medium were analyzed).10 If
an intermediate electronically similar to Boc-propargylamine
was chosen and reacted with 9, the high regioselectivity
experienced with the aforementioned reaction may be
translated to an alternative system. The pyrimidine 10 was
selected as a template to build the appropriate partner for a
reaction with 9 (Scheme 3).11 Starting material 10 was
protected with an acyl group12 and alkylated with propargyl
B
dx.doi.org/10.1021/op400137m | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
a
Scheme 4. SNAr addition of 3 and 7 to 13 under various conditions
a
Reagents and conditions: (a) 1.15 equiv of 3, DMA, TEA, or KHCO3, 20 or 46 °C, 12 h, >99.9:0.1 14/15; (b) 1.15 equiv of 3, DMA, TEA, or
KHCO3, 80 °C, 12 h, 91.3:8.7 14/15; (c) 1.15 equiv of 7, DMA, TEA, or KHCO3, 20 °C, 12 h, 80.5:19.5 16/17; (d) 1.15 equiv of 7, DMA, TEA, or
KHCO3, 80 °C, 12 h, 66.4:33.6 16/17.
a
Scheme 5. Initial GMP route to manufacture 1.
a
Reagents and conditions: (a) Ac2O, DMA, 130 °C, 3 h, 62%; (b) propargyl bromide (80% in PhMe), K2CO3, KHCO3, DMA, 30 °C, 24 h, 98%
conversion by LC; (c) 9, DMA, 30 °C, 10 h, 87% conversion by LC; (d) 3, DMA, 30 °C, 12 h, 97% conversion by LC; (e) CH3CN, HCl/IPA, 15
°C, 4 h, 57% from 11. (f) iPrOAc, H2O, K2CO3, 20 °C, then 7, iPrOAc, reflux, 4 h, 99% conversion by LC; (g) K2CO3, MeOH, 20 °C, 6 h, >99%
conversion by LC; (h) MtBE/iPrOAc, 25 to 10 °C, 58% yield from 14; (i) MtBE/iPrOAc, 25 to 10 °C, 92% yield, 98% purity.
bromide (80% in PhMe)13 to produce the desired cyclo-
addition partner 12. Addition of 9 to 12 at 40 °C readily
produced the desired isoxazole 13. Regioisomer 13a was not
detected in either the reaction medium or isolated product.
With a procedure in place to regioselectively form the desired
substituted isoxazole, the additions of 3 and 7 to 13 were
investigated (Scheme 4). An important advantage of proceed-
ing through 13 as compared to 4 was the symmetrical structure
of the 4,6-dichloropyrimidine core for addition reactions.
Judicious choice of reactant order and reaction conditions
could facilitate the selective addition of 3 or 7 to 13.
Development continued to focus on SNAr reactions because
of the broad knowledge gained during the initial scouting work
as well as the extensive literature precedent. Aromatic
metalation14 and transition metal-mediated reactions15 were
not investigated.
Reactivity/selectivity principles played a key role in
potentiating the SNAr addition of 3 and 7 to 13. The
acetoamino group provided enough electron-withdrawing
character to the pyrimidine ring (13) to promote the selective
monoaddition of 3 to 13 under mild reaction conditions
(Scheme 4, conditions (a)). Only at increased temperatures
(Scheme 4, conditions (b)) did the overaddition product with 3
appear (91.3:8.7 14/15).
C
dx.doi.org/10.1021/op400137m | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
a
Scheme 6. Alternative [3 + 2] cycloaddition conditions to form 13
a
Reagents and conditions: (a) cat. TEA, iPrOAc, reflux, 6 h, 92% yield.
In contrast, when 7 and 13 were reacted, overaddition of 7
occurred under both mild (Scheme 4, conditions (c), 80.5:19.5
16/17) and more forcing conditions (Scheme 4, conditions
(d), 66.4:33.6 16/17). Deacylated 13 was inactive to the SNAr
addition of 3, and only under the most forcing conditions
(K2CO3, DMA, 120 °C) was 7 added to deacylated 13 in low
conversion (11.1%). This collection of SNAr reaction data
demonstrated that the proper order of addition of the
nucleophiles (3 to 13, followed by 7) and selection of an
appropriately activated pyrimidine ring (11) can yield two
consecutive selective SNAr reactions.
With the major goal of demonstrating selective reactions
achieved, the end-game development was completed by
isolating 14 as an HCl salt. This control point allowed for
the removal of the unreacted reagents from the alkylation,
cycloaddition, and SNAr reactions. The final SNAr addition of 7
to the pyrimidine core and an acyl deprotection yielded 1 as
expected. With a suitable process route in place, scale-up
activities commenced.
First GMP Campaign. The process route and conditions
described in Scheme 5 were successfully employed in a small-
scale demonstration campaign that delivered 25 g of 1. A
campaign to deliver 100 g of 1 for IND enabling toxicology
studies using this route (Scheme 5) was also accomplished.
With the confidence gained from these two campaigns, the
route was scaled up and demonstrated in the first manufacture
of 1 under GMP. The yields and description presented below
are from that initial GMP campaign (Scheme 5).
with acceptable purity, but there were three aspects of the
synthesis that required optimization for future scale-up
campaigns. The synthesis of 9 was facile on scale for use in
the [3 + 2] cycloaddition reaction, but intermediate 9 was
found to have limited stability above 30 °C, which is
approximately the lowest temperature required for the
cycloaddition reaction to proceed. For the toxicology campaign,
92.6% conversion of 12 to 13 was achieved, and during the
initial GMP campaign, only 86.8% conversion was realized,
both using 2.1 equiv of 9.17
Intermediate 14·HCl was effective as a short-term isolation
point, but the long-term stability of this salt was not as robust as
initially anticipated. After one month of storage at ambient
temperature, a sample of 14·HCl from the initial GMP
campaign partially deliquesced to a mixture of deacylated 14
and numerous small, unidentified impurities (starting purity =
83.8%,18 purity at one month time point = 72.9%). A similar
phenomenon was seen with development and toxicology
batches (LC purity ≥ 91.0%) after ∼4 months. The lower
conversion of 12 to 13 and subsequent lower purity of 14·HCl
realized from the initial GMP campaign may be responsible for
the decreased stability of 14·HCl when compared to that of 14·
HCl samples from previous campaigns. Anticipating that similar
purities of 14·HCl may be obtained in future GMP campaigns
and hold times of 14·HCl may increase during multibatch
processing, an alternative counterion for 14 or a different
control point in the second GMP synthesis would need
implementation.
The acylation of 10 proceeded smoothly, providing 11 in
modest yield and high purity. The alkylation of 11, [3 + 2]
cycloaddition of 9 and 12, and SNAr addition of 3 to 13 were
telescoped by employing DMA as the solvent for all three
reactions. The bases for the three telescoped reactions, K2CO3
(freshly milled prior to charge) and KHCO3, were charged
prior to the initial alkylation reaction to minimize the number
of solid charging manipulations. The alkylation occurred
without incident although reaction times were longer on scale
(24 vs 11 h) when compared to the toxicology campaign.
Intermediate 9 was synthesized in a separate reactor using the
conditions outlined in Scheme 2 and was successfully added
over 2 h to 12 to provide 13 as a single regioisomer.
The SNAr addition of 3 to 13 was successful under mild
conditions, and the resulting product, 14, was isolated as an
HCl salt. Intermediate 14·HCl was converted to the free base, 7
was added to 14, the acetoamino group was removed, and 1
was crystallized to 96.5% purity from a 30-vol 2:1 MtBE/
iPrOAc solution. A subsequent 17-vol 1.4:1 MtBE/iPrOAc
crystallization provided 4.59 kg of 1 in 92% yield and 98%
purity (18.4% yield from 10).16
Finally, 1 was isolated as a mixture of an anhydrous and a
hydrated crystal form. The physiochemical properties of the
anhydrous form were far superior to the hydrate and other
known forms. The anhydrous form was not hygroscopic and
had the highest melting point (152 °C) of all forms isolated
(115 °C for the hydrate, other forms had melting points below
125 °C). The delivery of 1 as a single, anhydrous crystal form
was desirable to ensure a robust stability profile upon storage
and a consistent dissolution performance during drug product
formulation.
Second-Generation Process Development. The con-
cerns of the first GMP batch were addressed through
development work prior to the initiation of the second GMP
campaign. The formation of the isoxazole subunit on 13 was
successfully switched to an analogous 1,3-dipolar cycloaddition
procedure using 1,4-phenylene diisocyanate, 2-methyl-1-nitro-
propane (∼$1000/kilo, 1−3 week lead time), and catalytic
TEA in iPrOAc under reflux conditions (Scheme 6).19
For this key reaction, 12, the diisocyanate, base, and solvent
were charged together, and the mixture was heated at reflux
temperature. The 2-methyl-1-nitropropane was dosed to the
reaction mixture over ∼30 min, and heating was continued
until the desired reaction end point was reached. The 1,4-
First GMP Campaign Assessment. The first GMP
synthesis was capable of delivering the desired amount of 1
D
dx.doi.org/10.1021/op400137m | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
a
Scheme 7. Second GMP campaign to manufacture 1
a
Reagents and conditions: (a) Ac2O, DMA, 130 °C, 10 h, 71%; (b) propargyl bromide (80% in PhMe), K2CO3, KHCO3, DMA, 30 °C, 15 h, 65%
yield; (c) 1,4-phenylene diisocyanate, TEA, 2-methyl-1-nitropropane, iPrOAc, 80 °C, 6 h, 99% conversion by LC; (d) 3, KHCO3, DMA, 38 °C, 13 h,
98% conversion by LC; (e) 7, iPrOAc, 90 °C, 255 min, >99% conversion by LC; (f) K2CO3, 20 °C, 12 h, >99% conversion by LC; (g) iPrOAc, 60 to
20 °C, 24 h then −10 °C, 12 h, 40% yield from 12, 98% purity; (h) iPrOAc/EtOH, 53 to 2 °C, 86% yield, 16.4% from 10, 99% purity.
phenylene diisocyanate-derived polymers were easily filtered
away from the reaction solution upon cooling and water
addition. This alternative cycloaddition reaction did not require
a separate vessel to form the cycloaddition partner (9), and
triethylamine and unreacted 2-methyl-1-nitropropane were
innocuous in the downstream chemistry. Importantly, 13a
was not detected in the cycloaddition reaction during
development. A safety assessment of this reaction using
reaction calorimetry revealed the process to be safe under the
conditions presented in the experimental details.20
Alkylation of 11 produced a crude LC reaction profile that
showed an almost 8-fold increase (10.6%) of an unidentified
impurity as compared to the initial GMP campaign (1.4%). The
reaction conditions between the two campaigns were
unchanged. LC/MS analysis of the enhanced impurity returned
a m/z+ of 433, a mass that matched the dimerized product 18
(Figure 1). This impurity was possibly the result of the
hydrolysis of 12 and subsequent SNAr addition of another
molecule of 12.
Investigations into a new volume-efficient crystallization
system that would provide 1 in high purity and as a single,
nonsolvated crystal form were also conducted. A crystallization
solvent screen with crude 1 revealed that numerous solvent
systems (iPrOAc and iPrOAc with MeOH, EtOH, or IPA)
were capable of improving the purity of crude 1 from as low as
80% to >97.0% purity. These systems also yielded the desired
anhydrous crystal form of 1 and an impurity profile that was
similar or better to the material obtained in the toxicological
and first GMP campaign. Crystallization and recrystallization of
1 in 10 vol of iPrOAc was thus chosen as the final isolation
method for the second GMP campaign.
Armed with the excellent impurity purging capability of the
10 vol iPrOAc crystallization and recrystallization, small-scale
(25-g) development batches proved that the control point of
14·HCl was not necessary, and the process could be telescoped
from isolated 12 to crude 1 successfully. After recrystallization
of the crude material, 1 was isolated in improved purity
compared to the first GMP campaign (>99.0% vs 97.2%) and
with no new impurities above 0.1% when compared to the
material isolated from the toxicological and first GMP
campaign.
Figure 1. Impurity formed during the alkylation of 11.
Examination of historic in-process control chromatograms
revealed that the impurity 18 was present in all alkylation
reactions. The toxicology campaign and the initial GMP
campaign contained 1.1% and 1.4% of 18, respectively. All
route scouting and development reactions had 18 at levels of
0.2% to 1.5% in the final alkylation reaction solutions and in
examples where crude 18 was precipitated from a 10−15 °C
water addition to the bulk DMA solution. LC/MS analyses of 1
manufactured for the toxicology and first GMP campaign
revealed that 18 was not detected.
To challenge the newly implemented iPrOAc crystallization
and recrystallization system, a small portion (500 mL) of bulk
reaction solution containing 12 was removed, diluted with
water, and extracted with MEK, and the desired product was
precipitated from a room temperature 1.3:1.0 water/IPA
solution. Isolated intermediate 12, containing 12.0% 18, was
carried through the process (Scheme 7), and the new iPrOAc
crystallization and recrystallization were utilized. Crude 1 was
readily purified >99.0% with impurity 18 not detected, and
Second GMP Campaign. The second GMP campaign was
run on a 26-kg scale based on 10 and is described in Scheme 7.
The acyl protection of 10 proceeded as previously described,
but the yield was improved by employing 1:1 IPA/H2O as the
antisolvent instead of water to precipitate 11.
E
dx.doi.org/10.1021/op400137m | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
impurity 19 (Figure 2) below 0.1%. The GMP campaign was
thus continued.
and practical cycloaddition procedure utilizing 12, 1,4-phenyl-
enediisocyanate, catalytic TEA, and 2-methyl-1-nitropropane to
synthesize 13 was successfully introduced at the second GMP
campaign. Both GMP campaigns used reactivity/selectivity
principles to add 3 by SNAr to a symmetrical dichloropyr-
imidine (13) without overaddition. A volume-efficient crystal-
lization/recrystallization procedure was also developed to yield
the most stable crystal form of 1. The final process was capable
of producing 11.24 kg (16% yield) of 1 with a 99.35% LC
purity.
EXPERIMENTAL SECTION
■
General. An HP 1110 liquid chromatograph equipped with
a photodiode array detector was used for in-process analyses
and final analytical release. The analysis method used a 32-min
method ramping from 85:15 to 35:65 (A:B; solvent A = 10 mM
borate buffer, pH = 8.65 in 95:5 DI water/ACN; solvent B =
ACN) over 25 min, and 35:65 to 5:95 (A:B) from 25 to 32 min
with a Gemini C-18, 150 mm × 4.6 mm, 5 μm column at 1.0
mL/min, column temperature of 25 °C, and detection at 235
nm. Approximate retention times (min): 1 (17.4), 3 (3.3), 10
(10.8), 11 (7.5), 12 (20.3), 13 (24.5), 14 (21.4). Reported
yields are uncorrected.
Figure 2. Impurity derived from 18.
Intermediate 12 was isolated as described above (18 =
12.0%) and submitted to the cycloaddition reaction yielding 13.
Regioisomer 13a was not detected. The two SNAr reactions, the
deprotection, and the initial crystallization to supply crude 1
also occurred smoothly, but the crude isolated 1 contained a
red tint. The red tint of the batch was attributed to the starting
material 3. While the same vendor was used to supply 3 for the
toxicology and the first and second GMP campaigns, the lot
used in the second GMP campaign was a red oil, while 3 used
in all previous development work was a yellow oil.
Small-scale test crystallizations with crude 1 showed that the
originally planned iPrOAc recrystallization was not capable of
removing the red tint. Changing the final purification to a 40:1
iPrOAc/EtOH recrystallization successfully removed the red
tint and yielded 11.2 kg (16.4% from 10) of 1 in 99.35% purity
(18 = ND, 19 = 0.09%) as an off-white solid. DSC and XRPD
analyses confirmed that the desired single, anhydrous crystal
form was isolated.
N-(4,6-Dichloropyrimidin-2-yl)acetamide (11). A mix-
ture of 10 (26.56 kg, 161.9 mol) and DMA (90 L) was heated
to 130 °C and acetic anhydride (49.6 kg, 486 mol) charged
over 35 min. The reaction mixture was stirred at 130 °C for 10
h and cooled to 20 °C, and a 20 °C 1:1 IPA/H2O solution (264
L) was charged to the reaction mixture over 20 min. The
resulting suspension was stirred at 20 °C for 60 min and at 4 °C
for 135 min; the solids were collected, washed with a cold 1:1
IPA/H2O solution, and dried under reduced pressure at 45 °C
for 5 days to yield 23.65 kg (70.9% yield, 97.8% purity) of 11.
1H NMR (400 MHz, DMSO-d6): δ 11.12 (s, 1H), 7.58 (s, 1H),
2.15 (s, 3H). 13C NMR (100 MHz, DMSO-d6): δ 168.8, 161.3,
157.1, 115.4, 24.8. HRMS (ESI): [M + Na]+ m/z Calcd for
C6H5Cl2N3ONa 227.9707; Found 227.9703. Mp 190−197 °C
(dec.)
Control of Alkylation Impurity 18.21 A brief survey of
process changes to the formation of 12 (drying the milled
K2CO3, 10 °C reaction temperature, 10 reaction vol.) showed
that formation of 18 could be reduced to less than 1.9% by LC
analysis in numerous examples, but never eliminated.
Alternative bases (KH2PO4, K3PO4, NaHCO3, DBU,
DABCO, and 2,6-lutidine) were also examined, but impurity
18 was formed in all cases (≥0.9% by LC), except for 2,6-
lutidine, where only starting material was detected by LC
analysis. Rather than placing stringent water content controls
on the input materials, a robust recrystallization of crude 12
that was capable of eliminating 18 at levels encountered during
the second GMP campaign was investigated. A 2:1 H2O/IPA
recrystallization (14 vol H2O, 7 vol IPA, 50 °C, seed at 1:1
H2O/IPA), was found to be capable of purging up to 13.1% of
18 to <0.1% in a single attempt (74% yield). For samples of 12
that contained more than 13.1% of 18, the recrystallization
procedure could be repeated twice to reduce levels of 18 <
0.1%.
N-(4,6-Dichloropyrimidin-2-yl)-N-(prop-2-ynyl)-
acetamide (12). A mixture of 11 (23.63 kg, 114.7 mol), milled
potassium carbonate (19.86 kg, 143.4 mol), potassium
bicarbonate (22.96 kg, 229.4 mol), and DMA (72 L) was
heated to 30 °C. Propargyl bromide (80% in PhMe, stabilized
with MgO, 20.45 kg, 137.6 mol) was charged to the mixture
over 35 min and stirred at 30 °C for 15 h. CAUTION: Neat
propargyl bromide is shock sensitive and must be handled as
a stock solution (see reference 13). The reaction mixture was
cooled to 20 °C and diluted with water (483 L) and 2-
butanone (482 L), and the layers were separated, and the
organic phase was concentrated to 45 kg. The concentrated
organic solution was diluted with IPA (15.5 L) and water (20
L), and seeds of 12 were charged. The resulting suspension was
stirred at 20 °C for 50 min and 6 °C for 75 min, yielding 12 in
suspension. The solids were collected on a filter and washed
with two portions of cold 1:1 H2O/IPA (48 L), and 12 was
held in the filter at 25 °C under a stream of nitrogen until the
next step (18.1 kg, 64.7% isolated yield). An analytical sample
of 12 was prepared by a 2:1 H2O/IPA recrystallization to yield
CONCLUSION
■
Route development and multikilogram syntheses of 1 are
reported. The unselective reaction encountered between 4 and
5 in the medicinal chemistry synthesis was circumvented by the
implementation of a new process route designed around
pyrimidine 10. The substituted isoxazole unit was formed by a
regioselective [3 + 2] cycloaddition of 9 to 12 and was
demonstrated on a kiloscale GMP campaign. A more efficient
1
12 as pale-yellow flakes. H NMR (400 MHz, CDCl3): δ 7.12
(s, 1H), 4.87 (d, J = 2.4 Hz, 2H), 2.61 (s, 3H), 2.13 (t, J = 2.4
Hz, 1H). 13C NMR (100 MHz, CDCl3): δ 171.7, 162.0, 159.1,
115.9, 79.2, 70.8, 35.1, 27.0. HRMS (ESI): [M + H]+ m/z
F
dx.doi.org/10.1021/op400137m | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
Calcd for C9H8Cl2N3O 244.0044; Found 244.0035. Mp 64−65
°C.
under reduced pressure to 57 L and diluted with MeOH (93
L). The organic phase was charged with potassium carbonate
(27.5 kg, 198.7 mol) and stirred at 20 °C for 12 h (99.4%
conversion by LC). The solids were removed by filtration, the
filter cake was washed with iPrOAc (95 L), and the combined
organic solution was concentrated to 90 L. The organic
solution was diluted with water (100 L) and iPrOAc (100 L),
the layers were separated, and the aqueous phase was extracted
with iPrOAc (132 L). The combined organics were filtered
through a 1 μm filter, concentrated to 95 L, and seeded with
the anhydrous crystal form of 1. The mixture was stirred at 60
°C for 170 min, and allowed to cool to 20 °C over 24 h. The
solution was cooled to −10 °C over 12 h, the solids were
collected by filtration and washed with 151 L of cold iPrOAc to
yield crude 1 as a pale-red solid (13.0 kg). The isolated solids
were slurried with iPrOAc (120 L) and EtOH (3 L), heated to
53 °C for 2 h to dissolve the solids, cooled to 2 °C, and stirred
at 2 °C for 11 h. The solids were collected by filtration, rinsed
with cold iPrOAc (15 L), and dried at 20 °C under reduced
pressure for 72 h to yield 11.2 kg of 1 (86.2% yield, 16.4% from
10, 99.35% purity, 98.0% LC assay) as an off-white solid and
the anhydrous crystal form. 1H NMR (400 MHz, DMSO-d6): δ
11.92 (bs, 1H), 9.01 (bs, 1H), 6.88 (bs, 1H), 6.18 (s, 1H), 5.84
(bs, 1H), 5.73 (bs, 1H), 4.49 (d, J = 6.0 Hz, 2H), 3.38 (m, 4H),
2.93 (m, 1H), 2.30 (m, 4H), 2.17 (s, 3H), 1.80 (m, 1H), 1.16
(d, J = 7.2 Hz, 6H), 0.85 (m, 2H), 0.62 (m, 2H). 13C NMR
(100 MHz, DMSO-d6): δ 171.9, 168.7, 163.4, 161.0, 160.9,
99.5, 75.9, 54.4, 45.9, 43.7, 37.0, 26.0, 21.6, 7.8. HRMS (ESI):
[M + H]+ m/z Calcd for C22H32N9O 438.2730; Found
438.2733. Mp 151−152 °C.
N-(4,6-Dichloropyrimidin-2-yl)-N-((3-isopropylisoxa-
zol-5-yl)methyl)acetamide (13). The solids held in the
previous step (18.1 kg, 74.2 mol) were dissolved with iPrOAc
(182 L) and mixed with 1,4-phenylenediisocyanate (35.67 kg,
222.7 mol) and triethylamine (0.60 kg, 5.9 mol). The mixture
was heated to 80 °C, 2-methyl-1-nitropropane (9.95 kg, 96.5
mol) was charged over 30 min, and the reaction mixture was
stirred at 80 °C for 6 h (98.7% conversion by LC). After
cooling to 20 °C, water (108 L) was charged to the mixture
followed by iPrOAc (56 L). The brown suspension was stirred
at 20 °C for 28 h and filtered, and the filter cake was washed
twice with iPrOAc (55 L) and twice with water (23 L). The
phases were separated, and the organic phase containing 13 was
concentrated to 90 L. An analytical sample of 13 was prepared
by concentrating 50 mL of the organic solution and submitting
the crude sample to chromatography (9:1 heptane/EtOAc to
1:1 heptane/EtOAc) to yield 13 as a white solid. 1H NMR (400
MHz, CDCl3): δ 7.10 (s, 1H), 6.03 (s, 1H), 5.37 (s, 2H), 2.97
(sep, J = 7.2 Hz, 1H), 2.61 (s, 3H), 1.22 (d, J = 7.2 Hz, 6H).
13C NMR (100 MHz, CDCl3): δ 172.0, 169.4, 167.8, 162.0,
159.3, 116.0, 101.0, 40.9, 26.9, 26.6, 21.7. HRMS (ESI): [M +
H]+ m/z Calcd for C13H15Cl2N4O2 329.0572; Found 329.0559.
Mp 54−55 °C.
N-(4-chloro-6-(5-cyclopropyl-1H-pyrazol-3-ylamino)-
pyrimidin-2-yl)-N-((3-isopropylisoxazol-5-yl)methyl)-
1
acetamide (14). The solution containing 13 (21.8 kg by H
NMR assay, 66.2 mol) was diluted with DMA (47 L), and 52 L
of solvent was removed by reduced pressure distillation. The
reaction mixture was charged with potassium bicarbonate (19.9
kg, 198 mol) and heated to 38 °C, and a solution of 3 (8.91 kg,
72.8 mol) in DMA (34 L) was charged to the bulk solution of
13 over 2 h. The reaction mixture was stirred at 38 °C for 11 h
(97.6% conversion by LC) and cooled to 20 °C. The reaction
mixture was filtered to remove inorganic salts and the filter cake
washed with iPrOAc (14 L). The reaction solution was held at
20 °C under a nitrogen atmosphere until the next processing
step. An analytical sample was prepared by extracting 14 from
100 mL of the reaction solution using iPrOAc and water, drying
the organic solution (Na2SO4), concentrating the organic layer,
and submitting crude 14 to chromatography (9:1 heptane/
EtOAc to 3:7 heptane/EtOAc) to yield 14 as a pale-orange oil.
1H NMR (400 MHz, CDCl3): δ 8.42 (bs, 1H), 7.27 (s, 1H),
6.78 (bs, 1H), 6.03 (s, 1H), 5.88 (s, 1H), 5.30 (s, 2H), 2.99
(sep, J = 7.2 Hz, 1H), 2.54 (s, 3H), 1.86 (m, 1H), 1.23 (d, J =
7.2 Hz, 6H), 0.98 (m, 2H), 0.72 (m, 2H). 13C NMR (100
MHz, CDCl3): δ 172.7, 169.6, 168.9, 161.2, 159.8, 159.3, 148.3,
147.1, 100.9, 100.3, 93.5, 77.4, 41.4, 26.6, 26.3, 21.7, 8.1, 7.2.
HRMS (ESI): [M + H]+ m/z Calcd for C19H23ClN7O2
416.1602; Found 416.1590.
ASSOCIATED CONTENT
■
S
* Supporting Information
1H and 13C NMR spectra of 1, 11, 12, 13, and 14, a DSC
thermogram of the anhydrous form of 1, and reaction
calorimetry data for the conversion of 12 to 13 using 2-
methyl-1-nitropropane. This material is available free of charge
via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Author
*E-mail: nguz@exelixis.com. Telephone: +1-650-837-7461.
Present Addresses
‡Presently at Pharmacyclics, Inc., 999 East Arques Avenue,
Sunnyvale, CA 94085, United States
⊥Presently at Novartis Pharma AG, Novartis Campus, CH-
4002, Basel, Switzerland
Notes
The authors declare no competing financial interest.
N4-(5-Cyclopropyl-1H-pyrazol-3-yl)-N2-((3-isopropyli-
soxazol-5-yl)methyl)-6-(4-methylpiperazin-1-yl)-
pyrimidine-2,4-diamine (1). The reaction mixture contain-
ing 14 (66.2 mol) from the previous step was heated to 88 °C,
7 (19.4 kg, 198.7 mol) was charged over 15 min, and the
resulting mixture was heated at 90 °C for 255 min (99.8%
conversion by LC). After cooling to 20 °C, iPrOAc (159 L) and
H2O (159 L) were charged to the reaction solution, the layers
were separated, and the aqueous phase was extracted with
iPrOAc (162 L). The combined organic layers were
concentrated under reduced pressure to 57 L and diluted
with MeOH (93 L). The organic phase was concentrated again
ACKNOWLEDGMENTS
■
We acknowledge Anson Fatland and James Robinson III for
their contributions to the early stages of the project. We thank
Zicheng Yang, Jing Yuan, Helen Liu, and Roger Sun for their
analytical development and support throughout the project. We
also thank Sriram Naganathan, Norma Tom, Stephen Lau, and
Jo Ann Wilson for their review of the manuscript. High-
resolution mass spectrometry analyses were performed at the
Vincent Coates Foundation Mass Spectrometry Laboratory,
Stanford University.
G
dx.doi.org/10.1021/op400137m | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
NOESY experiments. The structure assignment of 9 was confirmed by
an X-ray crystal structure of 1.
REFERENCES
■
(1) (a) Catrina, S. B.; Lewitt, M.; Massambu, C.; Dricu, A.; Grunler,
̈
(11) A similar strategy to build an isoxazole attached to a pyrimidine
core is described in: (a) Reference 7a: (b) Kumar, C. H. V.; Kavitake,
S.; Kumar, S. S.; Cornwall, P.; Ashok, M.; Bhagat, S.; Manjunatha, S.
G.; Nambiar, S. Org. Process Res. Dev. 2012, 16, 1416−1421.
(12) The Boc protecting group was briefly investigated for the
second-generation approach but was not pursued.
(13) Propargyl bromide is shock sensitive as a neat liquid but is
generally considered safe to use as an 80% solution in PhMe. See
Bretherick’s Handbook of Reactive Chemical Hazards, 7th ed.; Elsevier:
Boston, 2007, Vol. 1.
J.; Axelson, M.; Biberfeld, P.; Brismark, K. Br. J. Cancer 2005, 92,
1467−1474. (b) Denly, A.; Wallace, J. C.; Cosgrove, L. J.; Forbes, B. E.
Horm. Metab. Res. 2003, 35, 778−785. (c) Dunn, S. E.; Ehrlich, M.;
Sharp, N. J.; Reiss, K.; Solomon, G.; Hawkins, R.; Baserga, R.; Barrett,
J. C. J. Cancer Res. 1998, 58, 3353−3361. (d) Hellawell, G. O.; Turner,
G. D. H.; Davies, D. R.; Poulsom, R.; Brewster, S. F.; Macaulay, V. M.
Cancer Res. 2002, 62, 2942−2950. (e) Lahm, H.; Amstad, P.; Wyniger,
J.; Yilmaz, A.; Fischer, J. R.; Schreyer, M.; Givel, J. C. Int. J. Cancer
1994, 58, 452−459. (f) Lee, C. T.; Park, K. H.; Adachi, Y.; Seol, J. Y.;
Yoo, C. G.; Kim, Y. M.; Han, S. K.; Shim, Y. S.; Coffee, K.; Dikov, M.
M.; Carbone, D. P. Cancer Gene Ther. 2003, 10, 57−63. (g) Mitsiades,
C. S.; Mitsiades, N. S.; McMullan, C. J.; Poulaki, V.; Shringarpure, R.;
Akiyama, M.; Hideshima, T.; Chauhan, D.; Joseph, M.; Libermann, T.
A.; Garcia-Echeverria, C.; Pearson, M. A.; Hofmann, F.; Anderson, K.
C.; Kung, A. L. Cancer Cell. 2004, 5, 221−230.
(14) (a) Bello, A. M.; Bende, T.; Wei, L.; Wang, X.; Majchrzak-Kita,
B.; Fish, E. N.; Kotra, L. P. J. Med. Chem. 2008, 51, 2734−2743.
(b) Turck, A.; Ple,
57, 4489−4505.
́ ́
N.; Mongin, F.; Queguiner, G. Tetrahedron 2001,
(15) (a) Nowak, P.; Cole, D. C.; Brooijmans, N.; Bursavich, M. G.;
Curran, K. J.; Ellingboe, J. W.; Gibbons, J. J.; Hollander, I.; Hu, Y.;
Kaplan, J.; Malwitz, D. J.; Toral-Barza, L.; Verheijen, J. C.; Zask, A.;
Zhang, W.-G.; Yu, K. J. Med. Chem. 2009, 52, 7081−7089.
(b) Schomaker, J. M.; Delia, T. J. J. Org. Chem. 2001, 66, 7125−
7128. (c) Peng, Z.-.H.; Journet, M.; Humphrey, G. Org. Lett. 2006, 8,
395−398. (d) Yaziji, V.; Coelho, A.; Maatougui, A. E.; Brea, J.; Loza,
M. I.; Garcia-Mera, X.; Sotelo, E. J. Comb. Chem. 2009, 11, 519−522.
(16) The initial drug product formulation was a 50 mg/mL, 10 mL
sterile solution of 1. Delivering the freebase of 1 provided the latitude
for the formulations development scientists to explore as many
counterions and dissolution aids as possible.
(2) (a) Summy, J. M.; Gallick, G. E. Cancer Metastasis Rev. 2003, 22,
337. (b) Yeatman, T. J. Nat. Rev. Cancer 2004, 4, 470−480.
(3) (a) An, X.; Tiwari, A. K.; Sun, Y.; Ding, P.-R.; Ashby, C. R.; Chen,
Z.-S. Leukemia Res. 2010, 34, 1255−1268. (b) Hantschel, O.; Superti-
Furga, G. Nat. Rev. Mol. Cell Biol. 2004, 5, 33−44.
(4) Nicolini, F. E.; Mauro, M. J.; Martinelli, G.; Kim, D.-W.; Soverini,
S.; Muller, M.; Hochhaus, A.; Cortes, J.; Chuah, C.; Dufva, I. H.;
̈
Apperley, J. F.; Yagasaki, F.; Pearson, J. D.; Peter, S.; Sanz Rodriguez,
C.; Preudhomme, C.; Giles, F.; Goldman, J. M.; Zhou, W. Blood 2009,
114, 5271−5278.
(5) Chen, J.; Dalrymple, L. E.; Epshteyn, S.; Forsyth, T. P.; Huynh,
T. P.; Ibrahim, M. A.; Leahy, J. W.; Lewis, G. L.; Mann, G.; Mann, L.
W.; Noguchi, R. T.; Ridgway, B. H.; Sangalang, J. C.; Schnepp, K. L.;
Shi, X.; Takeuchi, C. S.; Williams, M. A.; Nuss, J.; Cheung, A. K. PCT
Int. Appl. WO/2006/074057, 2006. Chem Abstr. 145, 137879.
(6) (a) Brown, D. J. In The Pyrimidines; Interscience Publishers: New
York, 1994. (b) Littke, A. F.; Fu, G. C. Angew. Chem., Int. Ed. 2002, 41,
4176−4211. (c) Parent, A. A.; Gunther, J. R.; Katzenellenbogen, J. A. J.
Med. Chem. 2008, 51, 6512−6530. (d) Padilla, A. G.; Pearlman, B. A.
Org. Process Res. Dev. 2006, 10, 921−926. (e) Delia, T. J.; Stark, D.;
Glenn, S. K. J. Heterocycl. Chem. 1995, 32, 1177−1180. (f) Delia, T. J.;
Nagarajan, A. J. Heterocycl. Chem. 1998, 35, 269−273. (g) Joshi, S.;
Maikap, G. C.; Titirmare, S.; Chaudhari, A.; Gurjar, M. K. Org. Process
Res. Dev. 2010, 14, 657−660. (h) Sagi, V. N.; Liu, T.; Lu, X.; Bartfai,
T.; Roberts, E. Bioorg. Med. Chem. Lett. 2011, 11, 7210−7215.
(i) Zanda, M.; Talaga, P.; Wagner, A.; Mioskowski, C. Tetrahedron
Lett. 2000, 41, 1757−1761.
(17) When over 3 equiv of 9 was charged to push the conversion of
12, the subsequent SNAr addition reaction time between 3 to 13 was
nearly 2-fold, and 14·HCl precipitated as a difficult-to-filter waxy solid.
(18) The intermediate 14·HCl contained ∼13% of the SNAr product
between unreacted 3 and 12.
(19) (a) Kantorowski, E. J; Brown, S. P.; Kurth, M. J. J. Org. Chem.
1998, 63, 5272−5274. (b) Mukaiyama, T.; Hoshino, T. J. Am. Chem.
Soc. 1960, 82, 5339−5342. (c) A related system: Cecchi, L.; De Sarlo,
F.; Marchetti, F. Eur. J. Org. Chem. 2006, 4852−4860.
(20) Reaction calorimetry using an RC-1 instrument returned a DSC
onset 160 °C, −138 J/g. The reaction data are presented in the
Supporting Information.
(21) The root cause for the nearly 8-fold increase in 18 was
investigated shortly after the completion of the second GMP
campaign. The alkylation step of this campaign was presumably run
with water levels higher than were encountered during development
and the previous scale-up campaign, likely causing the formation of 18.
Water content analysis of 12, propargyl bromide, and the reaction
solvent from the second campaign showed values similar to those in
the two development batches and the first GMP campaign. Potential
sources of water introduction pointed to the absorption of moisture on
the postmilled K2CO3 or a wet feed line.
(7) (a) Nowak, T.; Purkiss, S. C.; Thomas, A. P. PCT Int. Appl.
WO/2008/117051, 2008. Chem. Abstr. 149, 425969. (b) Wang, T.;
Lamb, M. L.; Scott, D. A.; Wang, H.; Block, M. H.; Lyne, P. D.; Lee, J.
W.; Davies, A. M.; Zhang, H.-J.; Zhu, Y.; Gu, F.; Han, Y.; Wang, B.;
Mohr, P. J.; Kaus, R. J.; Josey, J. A.; Hoffmann, E.; Thress, K.;
MacIntyre, T.; Wang, H.; Omer, C. A.; Yu, D. J. Med. Chem. 2008, 51,
4672−4684.
(8) O’Brien, D. E.; Weinstock, L. T.; Cheng, C. C. J. Med. Chem.
1968, 11, 387−388.
(9) (a) Liu, K.-C.; Shelton, B. R.; Howe, R. K. J. Org. Chem. 1980, 45,
3916−3918;(b) Barlaam, B.; Pape, A.; Thomas, A. PCT Int. Appl.
WO/2003/048133, 2003. Chem. Abstr. 139, 36535. (c) Barbachyn, M.
R.; Cleek, G. J.; Dolak, L. A.; Garmon, S. A.; Morris, J.; Seest, E. P.;
Thomas, R. C.; Toops, D. S.; Watt, W.; Wishka, D. G.; Ford, C. W.;
Zurenko, G. E.; Hamel, J. C.; Schaadt, R. D.; Stapert, D.; Yagi, B. H.;
Adams, W. J.; Friis, J. M.; Slatter, J. G.; Sams, J. P.; Oien, N. L.; Zaya,
M. J.; Weinkers, L. C.; Wynalda, M. A. J. Med. Chem. 2003, 46, 284−
302. (d) Girardin, M.; Dolman, S. J.; Lauzon, S.; Ouellet, S. G.;
Hughes, G.; Fernandez, P.; Zhou, G.; O’Shea, P. D. Org. Process Res.
Dev. 2011, 15, 1073−1080.
(10) Although the regioisomer of 9 was not available at the time of
process development, LC/MS analyses of 9 and subsequent
intermediates found no evidence of regioisomers derived from 9.
The chemical structure of 9 was initially assigned by COSY and
H
dx.doi.org/10.1021/op400137m | Org. Process Res. Dev. XXXX, XXX, XXX−XXX