Multi-kilo delivery of AMG 925 featuring a buchwald-hartwig amination and processing with insoluble synthetic intermediates

Article abstract of DOI:10.1021/op500367p

The development of a synthetic route to manufacture the drug candidate AMG 925 on kilogram scale is reported herein. The hydrochloride salt of AMG 925 was prepared in 23% overall yield over eight steps from commercially available raw materials, and more than 8 kg of the target molecule were delivered. The synthetic route features a Buchwald-Hartwig amination using BrettPhos as ligand and conducted to afford 12 kg of product in a single batch. In addition, this work highlights the challenges associated with the use of poorly soluble process intermediates in the manufacture of active pharmaceutical ingredients. Creative solutions had to be devised to conduct seemingly routine activities such as salt removal, pH adjustment, and heavy metal scavenging due to the low solubility of the process intermediates. Finally, a slurry-to-slurry amidation protocol was optimized to allow for successful scale-up.

Full text of DOI:10.1021/op500367p

Communication
pubs.acs.org/OPRD
Multi-Kilo Delivery of AMG 925 Featuring a Buchwald−Hartwig
Amination and Processing with Insoluble Synthetic Intermediates
Caroline Affouard,‡ Richard D. Crockett,† Khalid Diker,‡ Robert P. Farrell,† Gilles Gorins,‡
John R. Huckins,† and Seb Caille*,†
†Chemical Process R&D, Amgen Inc., One Amgen Center Drive, Thousand Oaks, California 91320
‡Norchim S.A.S., 33 Quai d’Amont, Saint Leu d’Esserent, France 60340
S
* Supporting Information
AMG 925 hydrochloride were necessary to support GLP
ABSTRACT: The development of a synthetic route to
manufacture the drug candidate AMG 925 on kilogram
scale is reported herein. The hydrochloride salt of AMG
925 was prepared in 23% overall yield over eight steps
from commercially available raw materials, and more than
8 kg of the target molecule were delivered. The synthetic
route features a Buchwald−Hartwig amination using
BrettPhos as ligand and conducted to afford 12 kg of
product in a single batch. In addition, this work highlights
the challenges associated with the use of poorly soluble
process intermediates in the manufacture of active
pharmaceutical ingredients. Creative solutions had to be
devised to conduct seemingly routine activities such as salt
removal, pH adjustment, and heavy metal scavenging due
to the low solubility of the process intermediates. Finally, a
slurry-to-slurry amidation protocol was optimized to allow
for successful scale-up.
toxicology studies and initiate clinical activities. Consequently,
the development of a robust process to deliver the first-in-
human drug substance batch for this program was undertaken.
RESULTS
■
i. Manufacture of Aminopyrimidine 1. The preparation
of eight kilograms of intermediate 1 was necessary to execute
the planned manufacturing campaign as per the reaction
sequence illustrated in Scheme 1. The route utilized to prepare
1 is shown in Scheme 2. None of the intermediates involved in
the synthetic sequence to manufacture 1 had solubility liabilities
that hampered processing. The procured raw material 5
contained 5 LCAP5 of dimer impurity 6. Fortunately,
purification of 5 was made possible using two aqueous lactic
acid washes, thus enabling the isolation of the starting material
(5) with only 0.9 LCAP of residual 6.
The manufacture of iodide 8 from 5 was carried out
successfully to afford 39 kg of product. Regioselective
deprotonation of 3-fluoropyridine (9) at the four position
was achieved using lithium tri-n-butylmagnesiate6 and the
lithium tripyridylmagnesiate thus generated was treated with
zinc bromide to afford the desired 4-pyridylzinc bromide
reagent (10). This material was utilized in situ in a Pd-catalyzed
Negishi cross-coupling with iodide 8.7 The formation of
regioisomer 12 was observed and represented ∼10% of the
mass balance. This side product was completely rejected during
the purification by crystallization of the target fluoropyridine
11. Overall, compound 11 was isolated in 54% corrected yield8
from 8 (20 kg of 11). Upon treatment with potassium t-
butoxide, 11 underwent cyclization to afford aminopyrimidine
1, and this material was isolated in high yield (85%) and purity
(99 LCAP, 99 wt %) after an activated charcoal treatment
followed by crystallization from dichloromethane and heptane.
ii. Solubility of Process Intermediates Used in the
Manufacture of AMG 925 HCl. In order to allow for
processing with high material throughput, it is generally
preferable to utilize solvents in which the synthetic
intermediates have a solubility of a 100 mg/mL or greater.9
To highlight the challenges driven by the poor solubility of the
intermediates shown in Scheme 1, solubility data in process
solvents typically used for API manufacture are presented in
INTRODUCTION
■
Clinical candidates and commercial drugs having planar,
heteroaromatic polycyclic structures are ubiquitous in the
pharmaceutical industry. One of the challenges involved in the
development of robust processes to manufacture these
materials is the low solubility of the synthetic intermediates
in most common solvents. In these cases, creative solutions
must be developed to address routine activities such as aqueous
extractions, pH adjustments, and minimization of entrainment
of starting material in the product. In the context of a recent
oncology program targeting dual inhibition of FLT31 and
CDK4,2 we developed an eight-step manufacturing process to
prepare a clinical candidate, AMG 925,3 as its hydrochloric acid
salt. The last three synthetic intermediates (Scheme 1) and the
drug substance AMG 925 hydrochloride (AMG 925 HCl) have
limited solubility in preferred process solvents. Examination of
the manufacturing route reveals the specific challenges to be
encountered considering the low solubility of the intermediates.
For example, elimination of residual palladium from compound
3 following a Buchwald−Hartwig amination4 and formation of
the free-base of compound 4 necessitate methods to keep these
molecules in solution in order to be successful. The amide
formation step to prepare AMG 925 is a slurry-to-slurry
transformation and effective process control is needed to avoid
starting material entrainment in the product. More than 5 kg of
Received: November 25, 2014
© XXXX American Chemical Society
A
DOI: 10.1021/op500367p
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
Scheme 1. Manufacturing route to AMG 925 hydrochloride
Scheme 2. Route to manufacture aminopyrimidine 1
Table 1. The poor solubility of these materials in most
preferred solvents (except acetic acid) is evident from
examination of the table and necessitates processing of slurry-
mixtures, use of large volumes of solvent (greater than 20 L/
kilogram of intermediate), and/or use of elevated temperatures
in order to complete the preparation of AMG 925 hydro-
chloride. As described below for the specific steps, this
processing challenge required creative solutions to ensure
robust drug substance manufacturing.
iii. Manufacture of Aminopyrimidine 3. The conversion
of 1 to aminopyrimidine 3 (Scheme 1) without the use of
transition-metal catalysis was investigated10 and very low yields
(<10%) of desired product 3 were observed. Several sources of
palladium were found to be competent in promoting this
transformation (>90% yields) and palladium acetate was
selected as a readily available precatalyst. Multiple ligands
could be used to conduct this Buchwald−Hartwig amination;
however, BrettPhos11 offered a decisively shorter processing
time (Table 2) and was selected for scale-up operations.
Noteworthy is the fact that tert-amyl alcohol was initially
utilized as solvent for ligand screening but was eventually
replaced with isopropyl alcohol as this alternative solvent
offered a slightly higher reaction rate at a lower temperature
(60 °C vs 100 °C). This is not surprising considering that the
necessary reduction of the precatalyst from Pd (II) to Pd (0)
occurs in less than 30 min at 20 °C using a β-hydrogen donor
B
DOI: 10.1021/op500367p
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
solubilization of the byproduct sodium chloride to prevent
blinding of the filter cloth during the filtration of 3, water (5
volumes) must be added to the reaction mixture prior to
filtration of the suspension. It is thus preferable to use a solvent
miscible with water at all compositions and temperatures, such
as isopropyl alcohol, to carry out this process and operate the
filtration of 3 from a monophasic liquid phase.
The target levels of heavy metal in the drug substance AMG
925 hydrochloride were <100 ppm, per regulatory guidance,12
much lower than the 700−800 ppm levels of palladium found
in crude 3 cakes. Considering the poor solubility of
downstream process intermediates and the limited knowledge
of downstream metal rejection, the specification for heavy metal
levels in intermediate 3 were set to <100 ppm. The removal of
palladium using aqueous N-acetylcysteine washes13 was not
possible in this case since 3 has insufficient solubility in solvents
that undergo phase separation with water. We consequently
turned our attention to the use of solid-supported palladium
scavengers and thus had to prepare a solution of 3 in reasonable
volumes of solvent (<15 volumes). This would allow for
palladium removal with the insoluble scavenger, filtration to
afford a solution of 3, and downstream crystallization of the
desired compound.
As can be seen in Table 1, mixtures of toluene and alcoholic
solvents uniquely offered high solubility for 3. The solubility of
3 in 80% (v/v) toluene/MeOH is 160 mg/mL at 50 °C,
allowing for both palladium scavenging using a solid scavenger
and subsequent removal by filtration of the scavenger at 50 °C
in only 10 volumes of solvent. After screening of potential
scavengers, silicyle−thiourea14 was selected since it adequately
reduced palladium levels. Metal scavenging was performed
using silicyle−thiourea (100 wt % relative to 3) in 80% (v/v)
toluene/MeOH and a total of 15 volumes of solvent (Scheme
3).15 Crude 3 was treated for 3 h, and the metal scavenger was
filtered. The resultant solution was seeded and cooled to 20 °C.
The crystallization was completed by addition of heptane, and
the slurry was filtered to afford product 3 containing <2 ppm of
residual Pd.
Table 1. Solubility data for AMG 925 process intermediates
(in mg/mL of solvent)
a
3
4
AMG 925
AMG 925 HCl
solvent
20 °C 60 °C 20 °C 50 °C 20 °C 50 °C
MeOH
EtOH
IPA
0.4
0.3
0.5
<1
<1
3.7
2.2
1
2.4
2.5
<1
<1
<1
<1
<1
<1
<1
<1
2.4
7
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
2.5
1.2
2.2
ND
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
3.0
2.6
7.3
ND
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
ND
<1
<1
<1
<1
<2
<1
<1
<1
<1
<1
2.1
1.2
2.3
ND
CH3CN
H2O
toluene
THF
IPAc
acetone
MTBE
DMSO
DMF
<1
<1
1
3
NMP
18.4
21
25
ND
b
50% v/v toluene/
MeOH
80% v/v toluene/
MeOH
43
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
80% v/v toluene/
MeOH (at 40 °C)
80% v/v toluene/
MeOH (at 50 °C)
120
160
50% v/v toluene/EtOH 23
80% v/v toluene/EtOH 38
ND
ND
47
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
NMP (at 80 °C)
ND
60% NMP/H2O (at
80 °C)
ND
30
50% NMP/H2O (at
80 °C)
ND
10
ND
ND
ND
ND
AcOH
49.5
ND
ND
ND
ND
ND
ND
ND
ND
ND
66
150
ND
ND
73
13
21
26
48
70
86
25% v/v AcOH/H2O
30% v/v AcOH/H2O
50% v/v AcOH/H2O
70% v/v AcOH/H2O
ND
ND
ND
ND
ND
ND
42
195
62
a
Assayed by quantitative HPLC analysis of the filtrate obtained by
agitation of a suspension of solute (in excess of soluble amount) in
b
specified solvent for a minimum of 24 h at the listed temperature. ND
iv. Manufacture of Amine 4. The t-butylcarbamate
(BOC) cleavage process used to generate amine 4 presented
challenges driven by the solubility of the desired product. The
hydrolysis process to generate 4 required the use of multiple
equivalents of acid in order to go to completion, even at
elevated temperature (70−100 °C). Protonation of one or
several of the basic nitrogen sites on 3 coupled with the poor
solubility of the salts thus generated is the cause of this problem
as demonstrated by chlorine content and HPLC analyses of
reaction mixture solids isolated from this process. Additionally,
the several equivalents of acid used in the deprotection process
had to be subsequently quenched with an equimolar amount of
base, thus generating large amounts of salt that needed to be
separated from 4 prior to the next step.16 Finally, product 4 did
not have sufficient solubility (see Table 1) in solvents that are
immiscible with water; thus, the salt byproducts could not be
simply removed by aqueous extraction.
= not determined.
a
Table 2. Ligand screen for Buchwald−Hartwig amination
temp
time
(h)
b
entry
ligand
solvent
(°C)
% AY
3
% AY 1
1
2
3
4
5
6
7
BrettPhos
BrettPhos
XantPhos
XantPhos
XPhos
tAmOH
tAmOH
tAmOH
tAmOH
tAmOH
tAmOH
IPA
100
100
100
100
100
100
60
3.5
21
95.2
97.6
66.4
95.5
76.9
93.1
97.5
2.1
0.4
3.5
19
29.1
0.2
3.5
19.5
3.0
16.9
4.0
XPhos
BrettPhos
0.24
a
The reported experiments were performed in the presence of 1 mol
% Pd (OAc)2, 1.5 mol % ligand, 1.5 equiv of NaOtBu, 15 volumes of
solvent, and 1.1 equiv of 2. AY refers to assay yield.
b
As shown in Table 1, the solubility of this material in N-
methylpyrrolidinone (NMP, 47 mg/mL at 80 °C) is higher
than its solubility in alternative solvents. Moreover, the
solubility of 4 in 60% (v/v) NMP/H2O is 30 mg/mL, making
possible the following procedure to perform this process:
solvent such as isopropyl alcohol as determined using 31P NMR
experiments. One additional benefit to using isopropyl alcohol
as solvent for the amination is that this solvent is miscible with
water at all compositions, as opposed to tert-amyl alcohol.
Upon its formation, product 3 crystallizes from the reaction
mixture due to its low solubility in isopropyl alcohol (Table 1),
allowing direct isolation of the product. In order to ensure
(i) completion of the carbamate cleavage in NMP with the
required equivalents of acid at elevated temperature,
C
DOI: 10.1021/op500367p
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
Scheme 3. Manufacture of aminopyrimidine 3
Scheme 4. Manufacture of amine 4
(ii) quenching of the acid with an equimolar amount of base
added as an aqueous solution,
(iii) generation of a solution of 4 and the salts in NMP/H2O
at elevated temperature,
(iv) seeding of the crystallization with 4,
(v) charging of additional water as antisolvent at elevated
temperature,
(vi) cooling of the slurry of crystallized 4,
(vii) filtration to isolate compound 4 free of salts.
evaluated by accelerating reaction calorimetry and was deemed
safe to operate.20 Gas evolution of isobutylene and carbon
dioxide was experimentally measured to occur over the course
of >4 h, and the venting area available on the reaction train
employed was found to be sufficient to accommodate the needs
of the process on this scale.
A subsurface nitrogen sparge was performed on the NMP
solution of 3 prior to conducting the process in order to
remove adventitious oxygen from the reaction mixture. The
presence of oxygen in this case was found to cause the
formation of impurity 1321 from reaction of 4 with NMP
degradation product 5-hydroperoxy-1-methylpyrrolidin-2-one
(Scheme 4), a derivative of N-methylpyrrolidinone formed in
the presence of oxygen at high temperatures.22
Twenty-five volumes of NMP were utilized to perform the
process, and it was possible to achieve complete cleavage of the
t-butylcarbamate group at 80 °C using 6 equiv of HCl (charged
as 5 N isopropyl alcohol solution) in 15 h.17 After addition of
6.3 equiv of sodium hydroxide18 as a 10 N aqueous solution,
the pH was verified to be basic (pH 9−10). The solution was
seeded with 4, antisolvent water (10 volumes) was charged, and
the cooled slurry was filtered to afford salt-free desired material
4 in high corrected yield (95%) and purity (99.7 LCAP, 99.2 wt
%). To ensure process safety,19 the batch protocol was
v. Manufacture of AMG 925. As demonstrated in Table 1,
the amidation process affording AMG 925 from 4 needed to be
conducted as a slurry-to-slurry process in which the starting
material and the product have limited solubility if practical
solvent volumes were to be utilized for manufacture. The use of
D
DOI: 10.1021/op500367p
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
N-methylpyrrolidinone (NMP) as solvent for the process
maximizes this limited solubility. Different coupling reagents
were evaluated to prepare AMG 925 from 4 and glycolic acid.
Whereas the use of carbonyl diimidazole (CDI) or propyl
phosphonic anhydride (T3P) yielded the desired product in
less than 25% assay yield, utilizing diisopropylcarbodiimide
(DIC) in the presence of either hydroxybenzotriazole (HOBt)
or 2-cyano-2-(hydroxyimino)acetate (Oxyma) as catalyst
allowed formation of AMG 925 in greater than 95% assay
yield.23
Diisopropylcarbodiimide was selected as the preferred
reagent for this transformation due to its ease of handling
(nonviscous liquid) as well as the fact that the coupling
byproduct diisopropylurea has high solubility24 in NMP and
was thus completely rejected upon filtration of the product.
Oxyma25 was selected to promote this process since it has been
shown to have lower explosion risks than HOBt26 and that it
does not carry shipping restrictions.27 The main impurities
observed in this transformation were ester 1428 (Scheme 5)
Table 3. Equivalents of DIC and glycolic acid used in the
amidation process
a
b
glycolic acid
equiv
DIC
equiv
% AY AMG
925
entry
% AY 4 % AY 14
1
2
3
4
1.1
1.1
97.1
97.7
98.3
98.6
2.6
1.7
1.3
0.0
0.2
0.2
0.4
1.2
1.15
1.2
1.15
1.2
2.0
1.2
a
DIC was added to a mixture of 4, Oxyma (1.0 equiv), and glycolic
acid in NMP (15 volumes) at 60 °C, and the mixtures were agitated
b
for 2 h for these experiments. AY refers to assay yield.
a
Table 4. Temperature screen for the amidation process
b
entry
temperature (°C)
% AY AMG 925
% AY 4
% AY 14
1
2
3
4
25
50
55
60
68.3
95.4
97.0
98.3
29.1
3.3
0.2
0.3
0.3
0.2
2.1
1.3
a
DIC (1.2 equiv) was added to a mixture of 4, Oxyma (1.0 equiv), and
Scheme 5. Manufacture of AMG 925
glycolic acid (1.2 equiv) in NMP (15 volumes) at the designated
temperature and the mixtures were agitated for 3 h for these
experiments. AY refers to assay yield.
b
studied (25−60 °C), a slurry-to-slurry reaction is observed and
that after 3 h of proceeding at a given temperature, the reaction
mixture levels of 4 cannot be decreased by elevating the
temperature and/or providing additional equivalents of DIC
and glycolic acid. To corroborate these results, samples of the
reaction mixture at these different temperatures were analyzed,
and residual 4 was found to be entrained in the solid cakes of
product AMG 925. Consequently, we elected to conduct the
process by charging DIC to a slurry of 4, Oxyma, and glycolic
acid in NMP at 60 °C in order to ensure that the
transformation would occur within a narrow window ( 3
°C) of the target temperature.29
Finally, the amount of time used to charge DIC to the
amidation reaction mixture (4, glycolic acid, Oxyma, and NMP)
at 60 °C was evaluated with the objective of limiting starting
material entrainment. The results of this study are summarized
in Table 5. Since only a minor exotherm accompanies the
Table 5. Time of addition of DIC at 60 °C for the amidation
a
process
and starting material 4. Although these impurities are poorly
rejected upon isolation of AMG 925, 4 shows better rejection
(85% rejection starting with 1.5 LCAP) than 14 (20% rejection
starting with 0.9 LCAP) during the crystallization of the drug
substance AMG 925 hydrochloride. Consequently, the
equivalents of DIC and glycolic acid utilized in the amidation
process were screened with the primary objective of limiting the
levels of 14 generated during the course of the reaction (see
Table 3). The use of excess glycolic acid (two equivalents) led
to the formation of prohibitively large (1.2 LCAP) amounts of
ester 14, and the amidation conditions presented in entry 3
(Table 3) were selected as a reasonable compromise to carry
this transformation to completion and limit the amounts of 14
formed.
A temperature screen for the amidation process was
performed with the selected conditions, and the results are
presented in Table 4. The transformation rate is sluggish at 25
°C (entry 1), and in general the use of temperatures below 60
°C leads to unacceptably high levels of residual starting material
4. It is important to note that, within the temperature range
b
entry addition time at 60 °C % AY AMG 925 % AY 4 % AY 14
1
2
3
4
1 min
98.3
98.0
97.7
97.1
1.3
1.4
1.6
2.3
0.2
0.4
0.3
0.1
6 min
20 min
60 min
a
DIC (1.2 equiv) was added to a mixture of 4, Oxyma (1.0 equiv), and
glycolic acid (1.2 equiv) in NMP (15 volumes) at 60 °C for these
experiments. AY refers to assay yield.
b
addition of DIC over a period of 1 min (2 °C exotherm on
measured on a scale of 40 g of 4, 74 kJ/mol), the target
addition time for DIC was set to <20 min, and the actual
addition time in the pilot-plant was <2 min. The process was
conducted using 8 kg of 4 with success, and AMG 925 was
generated with less than 0.05 LCAP of either starting material 4
or amide 14. After cooling of the suspension to 20 °C,
methanol (5 volumes) was used as an antisolvent before
filtration30 of the batch.
E
DOI: 10.1021/op500367p
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
Scheme 6. Manufacture of AMG 925 HCl
vi. Manufacture of AMG 925 HCl. A hydrochloric acid
salt of AMG 925 was prepared for drug product development,
and this salt had very limited solubility in virtually all solvents
(Table 1). A water/acetic acid solvent mixture was selected for
the process since it offered a range of solubility values for the
hydrochloric acid salt that would allow for design of a
controlled crystallization. This selection was made despite the
fact that the use of acetic acid as solvent to crystallize a
molecule containing an alcohol group is clearly hampered by
potential formation of the corresponding acetate by Fischer
esterification. In this instance, 0.65 LCAP of ester 15 was
generated during the manufacture of AMG 925 hydrochloride.
This impurity was partially rejected and present in 0.25 LCAP
in the product cake.
A crystalline trihydrate polymorph of the drug substance was
determined to be the thermodynamically most stable form in
aqueous mixtures of ≥15% water, and this polymorph was
selected for the drug product formulation process. The salt was
crystallized using a 1:1 mixture of acetic acid and water
according to the protocol described in Scheme 6. Upon seeding
of a solution of AMG 925 hydrochloride at 45 °C, roughly 30%
of the material crystallized. After formation of this seed bed,
two heat cycles between 50 and 20 °C were performed over the
course of 12 h. These heat cycles successfully reduced the
number of fine (<10 μm) particles and improved the filtration
rate of the product on kilogram scale.31 After crystallization of
roughly 50% of the material at 20 °C, isopropyl alcohol was
added as an antisolvent in preparation for filtration of the batch.
The dynamic vapor sorption plot of AMG 925 hydrochloride
showed a constant water level corresponding to three
equivalents between 10% and 90% RH, and it was observed
experimentally that the material retained this level of hydration
upon drying with a humidified nitrogen stream having ≥10%
RH. Accordingly, the relative humidity of the drying nitrogen
stream utilized was maintained between 35% and 45% RH
during manufacture.
SUMMARY
■
A process to manufacture AMG 925 hydrochloride was
developed and demonstrated to produce 8 kg. The overall
yield over eight steps from commercially available amino-
pyrimidine 5 was 23%. The synthetic route features a
Buchwald−Hartwig amination using BrettPhos as ligand and
was conducted to afford 12 kg of product in a single batch. Due
to the low solubility of the synthetic intermediates in most
preferred solvents, processing had to be adjusted to facilitate
seemingly routine activities such as salt removal, pH adjust-
ment, and heavy metal scavenging. Finally, a slurry-to-slurry
amidation was optimized to allow for successful scale-up.
EXPERIMENTAL SECTION
■
M a n u f a c t u r e o f 5 - I o d o - N 4 - ( ( 1 R , 4 R ) - 4 -
methylcyclohexyl)pyrimidine-2,4-diamine (Iodide 8).
To a solution of 2-amino-4-chloropyrimidine (21.3 kg, 1
equiv, 156 mol) in n-butanol (170 L, 8 vol) were added finely
ground potassium bicarbonate (65.4 kg, 3.06 equiv., 468 mol)
followed by trans-4-methylcyclohexylamine (27.9 kg, 1.58
equiv., 246 mol). The mixture was heated to reflux (110−115
°C) and agitated for 10 h. The mixture was cooled to 20 °C.
Water (85 L, 4 vol) and MTBE (105 L, 4.9 vol) were added,
and the mixture was agitated at 20 °C for 45 min. The mixture
was filtered through Celite (7 kg, 33 wt %), and the Celite cake
was rinsed with MTBE (120 L, 5.7 vol). The phases were
separated, and the organic phase was washed with saturated
aqueous sodium chloride (42 L, 2 vol). The organic phase was
distilled at atmospheric pressure to generate a distillate volume
of 815 L while simultaneously adding heptane (730 L, 34 vol)
to maintain the vessel fill level at 140−225 L. The mixture was
cooled to 20 °C and agitated for 12 h. The mixture was filtered,
and the filter cake was washed with water (3 × 70 L, 3 × 3.4
vol). The wet solid was dried at 50 °C for 72 h under nitrogen.
The dried solid was dissolved in 2-methyltetrahydrofuran (160
L, 5.9 vol) and heated to 30 °C. N-iodosuccinimide (33.4 kg,
1.2 equiv) was added over a period of 2 h, and the mixture was
agitated at 30 °C for 12 h. The mixture was cooled to 20 °C,
F
DOI: 10.1021/op500367p
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
and aqueous 5N sodium hydroxide (76 kg, 3.1 equiv) was
added over 15 min at <25 °C. The biphasic mixture was
agitated for 15 min, and the phases were separated. The
aqueous layer was extracted with 2-methyltetrahydrofuran (26
L, 1 vol). The combined organic layers were washed with brine
(3 × 50 L, 3 × 1.9 vol). The organic layer was distilled at
reduced pressure (maximum temperature of 40 °C) to generate
a distillate volume of 85 L. Heptane (240 L, 9.1 vol) was added,
and the mixture was agitated at 20 °C for 1 h. The mixture was
filtered, and the filter cake was washed with heptane (2 × 36 L,
2 × 1.4 vol). The wet solid was dried under vacuum at 40 °C
for 22 h to afford 39.0 kg of iodide 8 in 72.2% yield (96.9
equiv) in eight portions over 1 h, while maintaining the
temperature of the mixture <40 °C. The mixture was heated to
70 °C and agitated for 3 h. An additional portion of tert-
butoxide (6.6 kg, 1.0 equiv) was added, and the mixture was
warmed to 75 °C and agitated for 3.5 h. The mixture was
cooled to 20 °C. Deionized water (54 L, 2.7 vol) and 2-
methyltetrahydrofuran (110 L, 5.5 vol) were added, and the
biphasic mixture was agitated for 15 min. The phases were
separated, and the aqueous layer was extracted with 2-
methyltetrahydrofuran (2 × 75 L, 2 × 4 vol). The combined
organic layers were filtered through a 0.25 μm cartridge and
washed with brine (100 L, 5.3 vol). The aqueous brine layer
was extracted with 2-methyltetrahydrofuran (75 L, 4 vol). The
combined organic layers were washed with brine (2 × 100 L, 2
× 5.3 vol). The organic phase was filtered through a 0.25 μm
cartridge. The organic phase was distilled at atmospheric
pressure and 80 °C to generate a distillate volume of 270 L.
Heptane (180 L, 9 vol) was added at 60 °C, and the suspension
was cooled to 20 °C. The suspension was filtered. The filter
cake was washed with heptane (2 × 40 L, 2 × 2 vol) and dried
at 50 °C for 24 h. The material was dissolved in dichloro-
methane (170 L, 10 vol), and activated charcoal (0.85 kg, 5 wt
%) was added. The mixture was agitated at 20 °C for 1 h and
filtered through a 0.25 μm cartridge. The vessel and filter line
were washed with dichloromethane (40 L, 2.2 vol). The
combined dichloromethane phases were distilled at atmos-
pheric pressure and 35 °C to generate a distillate volume of 195
L. Heptane (170 L, 10 vol) was added at 35 °C, and the
suspension was cooled at 20 °C. The suspension was filtered.
The filter cake was washed with heptane (2 × 40 L, 2 × 2 vol)
and dried at 50 °C for 48 h to afford 14.3 kg of
aminopyrimidine 1 in 84.9% yield (99.2 LCAP, 99.1 wt %).
1
LCAP, 95.6 wt %). Mp 138−140 °C; H NMR (300 MHz,
DMSO-d6) 7.79 (s, 1H), 6.08 (s, 2H), 5.40 (d, 1H, J = 12 Hz),
3.69−3.83 (m, 1H), 1.52−1.79 (m, 4H), 1.15−1.37 (m, 3H),
0.72−0.96 (m, 5H); 13C NMR (75 MHz, CDCl3) 162.5, 161.6,
159.4, 65.3, 49.9, 33.8, 33.0, 32.0, 22.2.
Manufacture of 5-(3-Fluoropyridin-4-yl)-N4-((1R,4R)-4-
methylcyclohexyl)pyrimidine-2,4-diamine (Fluoropyri-
dine 11). To a cold (−10 °C) suspension of magnesium
bromide (21.7 kg, 1.05 equiv) in tetrahydrofuran (230 L, 5.9
vol) was added n-butyllithium in hexane (99.6 kg, 23 wt %, 3.2
equiv) over 4 h at <−8 °C. 3-Fluoropyridine (32.9 kg, 3.0
equiv) was added to the reaction mixture over 5 h at <−8 °C.
To the reaction mixture was added a solution of zinc bromide
(76.5 kg, 3.0 equiv) in tetrahydrofuran (190 L, 4.9 vol) over 3 h
at <−2 °C. The reaction mixture was warmed to 15 °C over 1
h. A use-test was conducted using a sample of the mixture, a
sample of iodide 8, as well as a sample of tetrakis-
(triphenylphosphine) palladium, and the in-process control
test at the intermediate time point (2 h) showed expected
conversion. Consequently, iodide 8 (39.0 kg, 1.0 equiv) and
tetrakis(triphenylphosphine) palladium (5 kg, 0.04 equiv) were
added, and the mixture was heated to 65 °C. The mixture was
agitated for 5 h and cooled to 20 °C. An aqueous solution of N-
(2-hydroxyethyl)ethylendiamine-N,N′,N′-triacetic acid sodium
salt dihydrate (490 L, 10 wt %) was added to the reaction
mixture over 30 min. The biphasic mixture was agitated for 30
min. The layers were filtered and separated. An aqueous
solution of N-(2-hydroxyethyl)ethylendiamine-N,N′,N′-tria-
cetic acid sodium salt dihydrate (300 L, 5 wt %) was added
to the reaction mixture over 30 min. The biphasic mixture was
agitated for 30 min. The layers were separated. The organic
phase was distilled at atmospheric pressure to generate a
distillate volume of 500 L while simultaneously adding
isopropanol (120 L, 3 vol). During this time, the product
fluoropyridine 11 mostly precipitated out of solution. The
suspension was agitated at 20 °C for 30 min. An aqueous
solution of N-(2-hydroxyethyl)ethylendiamine-N,N′,N′-tria-
cetic acid sodium salt dihydrate (250 L, 5 wt %) was added
to the reaction mixture over 30 min. The mixture was agitated
for 1 h and filtered. The filter cake was washed with deionized
water (2 × 45 L) and dried at 50 °C for 48 h to afford 20.1 kg
of fluoropyridine 11 in 52.3% yield (99.3 LCAP, 93.0 wt %).
1
Mp 208−210 °C; H NMR (300 MHz, DMSO-d6) 8.43 (d,
1H, J = 3 Hz), 8.29−8.31 (m, 1H), 7.50 (s, 1H), 7.26−7.30 (m,
1H), 6.15 (s, 2H), 5.91 (d, 1H, J = 9 Hz), 3.82−3.96 (m, 1H),
1.51−1.71 (m, 4H), 1.08−1.28 (m, 3H), 0.75−0.95 (m, 5H).
Manufacture of tert-Butyl-2-((9-((1R,4R)-4-methylcy-
clohexyl)-9H-pyrido[4′,3′:4,5]pyrrolo[2,3-d]pyrimidin-2-
yl)amino)-7,8-dihydro-1,6-naphthyridine-6(5H)-carbox-
ylate (Aminopyrimidine 3). NaOtBu (2.99 kg, 31.1 mol, 1.2
equiv), 2 (7.67 kg, 28.5 mol, 1.1 equiv), 1 (7.30 kg, 25.9 mol,
1.0 equiv), and IPA (100 L, 13.6 vol.) were charged via a sealed
manhole to a 250 L vessel equipped with a reflux/return
condenser. The contents of the 250 L vessel were agitated, and
a subsurface N2 sparge was performed for 20 min. Pd(OAc)2
(0.058 kg, 0.26 mol, 0.01 equiv), BrettPhos (0.17 kg, 0.32 mol,
0.0125 equiv), and degassed32 IPA (10 L, 1.4 vol.) were
charged to a 10 L vessel under an atmosphere of N2. The
contents of the 10 L vessel were agitated for 0.5 h and charged
to the 250 L vessel. The contents of the 250 L vessel were
heated to 60 °C for 3 h. HPLC analysis of a process sample
showed 97.5 LCAP 3 and 0.24 LCAP 1. Degassed water (32.8
L, 4.5 vol.) was added to the contents of the 250 L reactor over
1 h. The contents of the 250 L vessel were agitated for 1 h at 60
°C, cooled to 20 °C over 2 h (linear ramp), and agitated for 12
h. The contents of the 250 L vessel were split in half and
filtered through two 24 in. Aurora filter-driers equipped with 12
μm PTFE filter cloths and under a positive pressure of nitrogen
(2−5 psig). The mother liquors were transferred to a 400 L
vessel. The 250 L vessel was rinsed with a solution of IPA (25.6
L, 3.5 vol.) and water (11.0 L, 1.5 vol.), and the rinse solution
was split in half to wash both process cakes. The wash solutions
were transferred to a 400 L vessel. The combined mother
1
Mp 140−142 °C; H NMR (300 MHz, DMSO-d6) 8.90−8.95
(m, 2H), 8.24 (d, 1H, J = 6 Hz), 7.82 (d, 1H, J = 6 Hz), 6.90 (s,
2H), 4.60−4.78 (m, 1H), 2.25−2.40 (m, 2H), 1.50−1.81 (m,
5H), 1.20−1.00 (m, 2H), 0.87 (d, 3H, J = 6 Hz).
Manufacture of 9-((1R,4R)-4-methylcyclohexyl)-9H-
pyrido[4′,3′:4,5]pyrrolo[2,3-d]pyrimidin-2-amine (Ami-
nopyrimidine 1). To a solution of fluoropyridine 11 (20.1
kg, 93 wt %, 1.0 equiv) in 1-methyl-2-pyrrolidinone (90 L, 4.5
vol) at 30 °C was added potassium tert-butoxide (20.0 kg, 3.0
G
DOI: 10.1021/op500367p
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
liquors and wash solutions were assayed for 3 losses (2.1%).
The filtration and wash process took approximately 2.5 h. The
cakes were dried under positive nitrogen pressure (dry
nitrogen, 0−5 psig) for 3 days. Cake samples showed weight
losses by TGA (up to 180 °C) of 0.19% and 0.27%. Metals
levels were measured in the cakes at 732−775 ppm Pd and
224−774 ppm of Na (ICP-MS analyses). Crude 3 (12.6 kg)
was packaged and charged via a sealed manhole to a 250 L
vessel equipped with a reflux/return condenser. Silicycle-
Thiourea (7.3 kg, 100 wt %) and a degassed22 solution of
toluene (81.6 L, 11.2 vol.) and MeOH (20.4 L, 2.8 vol.) were
charged. The contents of the 250 L vessel were heated to 50 °C
and agitated under an atmosphere of N2 for 3 h. The
supernatant Pd levels in the 250L vessel were measured to be
0.7 ppm. The contents of the 250 L vessel were warmed to 60
°C and filtered through a 14 in. Aurora filter-drier (jacket
temperature 60 °C) equipped with a 12 μm PTFE filter cloth
under a positive pressure of nitrogen (2−5 psig). The filtrate
was charged to a 400 L vessel with a jacket temperature set to
50 °C. The 250 L vessel was rinsed with a solution of toluene
(11.7 L, 1.6 vol.) and MeOH (2.9 L, 0.4 vol.), the silicycle−
thiourea cake was washed with this rinse solution, and the wash
solution was transferred to the 400 L vessel. The agitated
contents of the 400 L vessel were cooled to 35 °C under an
atmosphere of N2. A shaken slurry of 3 (0.133 kg, 0.25 mol,
0.018 equiv) in degassed heptane (1.8 L, 0.25 vol.) was charged
to the contents of the 400 L vessel. After a period of 10 min,
formation of a slurry in the 400 L vessel was visually observed.
The contents of the 400 L were agitated at 35 °C for 1 h and
cooled to 20 °C over 3 h (linear ramp). Degassed heptane (185
L, 25.4 vol.) was added to the contents of the 400 L vessel over
4 h (linear addition). The contents of the 400 L vessel were
agitated for 12 h, and the supernatant concentration of 3 was
measured by HPLC to be 1.3 mg/mL. The contents of the 400
L vessel were filtered through a 24 in. Aurora filter-drier
equipped with a 12 μm PTFE filter cloth under a positive
pressure of nitrogen (2−5 psig). The mother liquors were
transferred to a 250 L vessel. The 400 L vessel was rinsed with a
solution of toluene (11.0 L, 1.5 vol.), MeOH (2.9 L, 0.4 vol.),
and heptane (22.6 L, 3.1 vol.), the cake was washed with this
solution, and the wash solution was collected in a 250 L vessel.
The combined mother liquors and wash solutions were assayed
for 3 losses (2.4%). The filtration and wash process took
approximately 1 h. The cake was dried under positive nitrogen
pressure (dry nitrogen, 0−5 psig) for 2 days. A cake sample
showed a weight loss by TGA (up to 180 °C) of 2.1%, and the
material was packaged. Aminopyrimidine 3 was isolated in
90.6% yield (12.1 kg), 95.3% overall mass balance, 99 wt %, and
99.9 LCAP. The material contained 1 ppm of palladium. Mp
(11.0 kg, 99.0 wt %, 21.4 mol, 1.0 equiv) and degassed22 NMP
(275 L, 25 vol.) were charged via a sealed manhole to a 400 L
vessel equipped with a reflux/return condenser and a NaOH
scrubber. The contents of the 400 L vessel were agitated (100
rpm) and degassed using a subsurface N2 sparge (20 min). HCl
in IPA (21.1 kg, 5.56 M, 128.5 mol, 6.0 equiv) was charged.
The contents of the 400 L vessel were heated to 80 °C over 1.5
h and agitated for 16 h. HPLC analysis of a process sample
showed 99.7 LCAP 4 and 0.1 LCAP 3. A degassed aqueous
NaOH solution (17.8 kg, 10.0 N, 135.0 mol, 6.3 equiv) was
added over 45 min to the contents of the 400 L vessel.
Degassed water was added over 30 min (50 L, 4.5 vol.) to the
contents of the 400L vessel. The pH of the contents of the 400
L reactor was measured to be above 10. A shaken slurry of 4
(81 g, 0.21 mol, 0.01 equiv) in a degassed22 mixture of NMP
(0.55 L, 0.05 vol.) and water (0.22 L, 0.02 vol.) was charged to
the contents of the 400 L vessel. After a period of 10 min,
formation of a slurry in the 400 L vessel was visually observed,
and the contents of the 400 L were agitated at 80 °C for 30
min. Degassed water was added over 1 h (105 L, 9.5 vol.) to the
contents of the 400L vessel. The contents of the 400 L were
cooled to 20 °C over 8 h (linear ramp), agitated for 4 h, and the
supernatant concentration of 4 in the slurry was measured by
HPLC to be 0.4 mg/mL. The contents of the 400 L vessel were
filtered through a 24 in. Aurora filter-drier equipped with a 12
μm PTFE filter cloth under a positive pressure of nitrogen (2−
5 psig). The mother liquors were transferred to a 250 L vessel.
The 400 L vessel was rinsed with a solution of NMP (20.7 L,
1.9 vol.) and water (12.3 L, 1.1 vol.), the cake was washed with
this solution, and the wash solution was collected in a 250 L
vessel. The 400 L vessel was rinsed with water (33.0 L, 3.0
vol.), the cake was washed with this water, and the wash
solution was collected in a 250 L vessel. The last unit operation
(water rinse) was repeated twice. The cake was washed with
heptane (22.0 L, 2.0 vol.), and the wash solution was collected
in a 250 L vessel. The filtration and wash process took
approximately 16 h. The cake was dried under positive nitrogen
pressure (dry nitrogen, 0−5 psig) for 6 days. A cake sample
showed a weight loss by TGA (up to 180 °C) of 0.4%, and the
material was packaged. Amine 4 was isolated in 95.2% yield
(8.53 kg), 97.6% overall mass balance, 99.2 wt %, and 99.7
LCAP. Mp 276−278 °C; 1H NMR (400 MHz, AcOH-d4) 9.48
(d, 2H, J = 3 Hz), 8.72 (d, 1H, J = 6 Hz), 8.49−8.53 (m, 2H),
7.82 (d, 1H, J = 9 Hz), 4.85−4.96 (m, 1H), 4.55 (s, 2H), 3.74
(t, 2H, J = 6 Hz,), 3.26 (t, 2H, J = 7 Hz), 2.60−2.80 (m, 2H),
1.74 (br s, 1H), 1.28−1.40 (m, 2H), 1.07 (d, 3H, J = 7 Hz); 13C
NMR (100 MHz, AcOH-d4) 159.5, 159.3, 155.8, 152.4, 150.7,
139.3, 135.9, 134.5, 133.3, 127.8, 120.5, 117.5, 114.0, 107.3,
57.1, 44.7, 42.5, 35.2, 32.8, 30.8, 27.8, 22.6; exact mass
[C24H27N7 + H]+: calculated = 414.2406, measured =
414.2390.
1
250−252 °C; H NMR (400 MHz, CDCl3) 9.11 (s, 1H), 8.96
(s, 1H), 8.51 (d, 1H, J = 4 Hz), 8.39 (d, 1H, J = 8 Hz), 8.14−
8.25 (br s, 1H), 7.84 (d, 1H, J = 4 Hz), 7.48 (d, 1H, J = 8 Hz),
4.68−4.78 (m, 1H), 4.59 (s, 2H), 3.77 (t, 2H, J = 6 Hz), 2.93
(t, 2H, J = 6 Hz), 2.53−2.66 (m, 2H), 1.93−2.03 (m, 4H), 1.51
(s, 9H), 1.20−1.36 (m, 2H), 1.06 (d, 3H, J = 4 Hz); 13C NMR
(100 MHz, CDCl3) 158.0, 156.6, 154.8, 152.9, 152.0, 150.8,
141.0, 135.9, 134.8, 133.1, 129.9, 126.6, 114.0, 110.8, 106.6,
80.1, 54.6, 34.7, 32.1, 31.9, 30.2, 28.5, 22.3; exact mass
[C29H35N7O2 + H]+: calculated = 514.2930, measured =
514.2920.
Manufacture of 2-Hydroxy-1-(2-((9-((1R,4R)-4-methyl-
cyclohexyl)-9H-pyrido[4′,3′:4,5]pyrrolo[2,3-d]pyrimidin-
2-yl)amino)-7,8-dihydro-1,6-naphthyridin-6(5H)-yl)-
ethanone (AMG 925). Glycolic acid (1.76 kg, 23.1 mol, 1.2
equiv), Oxyma (2.73 kg, 19.3 mol, 1.0 equiv), 4 (7.96 kg, 19.3
mol, 1.0 equiv), and NMP (119 L, 15 vol.) were charged via a
sealed manhole to a 250L vessel equipped with a reflux/return
condenser. The contents of the 250 L vessel were agitated (130
rpm) under an atmosphere of N2 and heated to 60 °C. DIC
(2.91 L, 23.1 mol, 1.2 equiv) was added to the contents of the
250 L vessel over 1 min (5 °C exotherm was recorded), and the
addition line was rinsed with NMP (0.3 L, 0.04 vol., the rinse
Manufacture of 9-((1R,4R)-4-Methylcyclohexyl)-N-
(5,6,7,8-tetrahydro-1,6-naphthyridin-2-yl)-9H-pyrido-
[4′,3′:4,5]pyrrolo[2,3-d]pyrimidin-2-amine (Amine 4). 3
H
DOI: 10.1021/op500367p
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
solution was added to the contents of the 250 L vessel). The
contents of the 250 L vessel were agitated for 1.3 h. HPLC
analysis of a process sample showed 91.0 LCAP AMG 925 and
0.05 LCAP 4 (Oxyma accounts for 7.9 LCAP). The contents of
the 250 L were cooled to 20 °C over 2.5 h (linear ramp, 100
rpm agitation). MeOH (39.8 L, 5 vol.) was added to the
contents of the 250L vessel over 40 min (linear addition), and
the contents of 250 L vessel were agitated (60 rpm) for 12 h at
20 °C. The supernatant concentration of AMG 925 in the
slurry was measured by HPLC to be 2.2 mg/mL. The contents
of the 250 L vessel were filtered through a 24 in. Aurora filter-
drier equipped with a 12 μm PTFE filter cloth under a positive
pressure of nitrogen (2−5 psig). The mother liquors were
transferred to a 400 L vessel. The 250 L vessel was rinsed with
MeOH (59.7 L, 7.5 vol.), the cake was washed with this
MeOH, and the wash solution was collected in a 400 L vessel.
The last unit operation (MeOH rinse) was repeated. The
filtration and wash process took approximately 8 h. The cake
was dried under positive nitrogen pressure (dry nitrogen, 0−5
psig) for 36 h. A cake sample showed a weight loss by TGA (up
to 180 °C) of 1.0%, and the material was packaged. AMG 925
was isolated in 91.5% yield (8.31 kg), 95.5% overall mass
balance, 99.9 wt %, and 99.7 LCAP. Mp 213−215 °C; 1H NMR
(400 MHz, acetic acid-d4, mixture of two rotamers at 20 °C)
9.47−9.59 (m, 2H), 8.76 (d, 1H, J = 6 Hz), 8.55 (d, 1H, J = 6
Hz), 8.48 (d, 1H, J = 9 Hz), 7.79−7.92 (m, 1H), 4.95 (t, 1H, J
= 12 Hz), 4.87 and 4.68 (2 singlets, 2H), 4.47−4.59 (m, 2H),
4.04 and 3.80 (2 triplets, 2H, J = 6 Hz), 3.03−3.17 (m, 2H),
2.65−2.82 (m, 2H), 1.96−2.15 (m, 4H), 1.77 (br s, 1H), 1.39
(q, 2H, J = 12 Hz), 1.09 (d, 3H, J = 7 Hz); 13C NMR (100
MHz, acetic acid-d4, mixture of two rotamers at 20 °C) 171.9,
171.8, 158.4, 157.8, 154.7, 149.0, 148.9, 141.6, 135.2, 132.9,
126.3, 124.1, 123.6, 117.7, 113.7, 113.6, 107.5, 107.4, 60.1, 59.9,
56.3, 43.7, 42.6, 40.5, 38.7, 34.0, 31.5, 29.8, 28.8, 28.1, 21.5.
Manufacture of 2-Hydroxy-1-(2-((9-((1R,4R)-4-methyl-
cyclohexyl)-9H-pyrido[4′,3′:4,5]pyrrolo[2,3-d]pyrimidin-
2-yl)amino)-7,8-dihydro-1,6-naphthyridin-6(5H)-yl)-
ethanone Hydrochloride (AMG 925 HCl). AMG 925 (7.01
kg, 14.88 mol, 1.0 equiv) was charged via a sealed manhole
(due to the cytotoxic potential of the material) to a 250 L vessel
equipped with a reflux/return condenser. Acetic acid (35.0 L, 5
vol) and water (17.5 L, 2.5 vol) were charged to the 250 L
vessel. The contents of the 250 L vessel were heated to 55 °C
under an atmosphere of nitrogen and held for 8 h. Mechanical
agitation was initiated in the vessel (100 rpm), and the solution
was stirred for 1 h. The contents of the 250 L vessel were
passed through a 5 μm cartridge filter and transferred in a
preheated (50 °C) 400L reactor, equipped with a reflux/return
condenser. The 250 L vessel was rinsed with a solution of acetic
acid (7.0 L, 1.0 vol) and water (3.5 L, 0.5 vol), and the rinse
solution was transferred, through the cartridge filter, in the 400
L vessel. To the mechanically agitated (100 rpm) content of the
400 L vessel at 50 °C and under an atmosphere of nitrogen
were added an aqueous 1 N HCl solution (15.6 L, 15.6 mol,
1.05 equiv) over 10 min and water (4.6 L, 0.65 vol) over 5 min.
The contents of the 400L reactor were cooled to 45 °C and the
agitation rate was reduced (25 rpm). A shaken slurry of AMG
925 HCl (240 g, 0.42 mol, 0.028 equiv., unmilled, D10 = 3.9
μm, D50 = 15.5 μm, D90 = 35.3 μm, VM= 18.1 μm) in IPA (0.88
L, 0.125 vol) and water (0.88 L, 0.125 vol) was charged and
after a period of 10 min, formation of a slurry in the 400 L
vessel was visually observed. The contents of the 400 L vessel
were agitated at 45 °C for 1 h, cooled to 25 °C over a period of
1 h (linear cooling ramp), and heated again to 50 °C. The
contents of the 400 L vessel were agitated at 50 °C for 4 h,
cooled to 20 °C over a period of 1 h (linear cooling ramp), and
agitated at 20 °C for 4 h. The concentration of AMG 925 HCl
in solution was measured by HPLC to be 53.4 mg/mL. The
contents of the 400 L vessel were heated to 50 °C, agitated for
4 h, cooled to 20 °C over a period of 1 h (linear cooling ramp),
and agitated at 20 °C for 1 h. The concentration of AMG 925
HCl in solution was measured by HPLC to be 60.5 mg/mL.
IPA (35.0 L, 5.0 vol) was added to the contents of the 400 L
vessel over 2.5 h (linear addition ramp). The agitation rate was
increased (35 rpm), and IPA (77.0 L, 11.0 vol) was added to
the contents of the 400 L vessel over a period of 2.5 h (linear
addition ramp). The contents of the 400 L vessel were agitated
for 12 h (25 rpm), and the supernatant concentration of AMG
925 HCl was measured by HPLC to be 2.4 mg/mL. The
contents of the 400 L vessel were filtered through a 24 in.
Aurora filter-drier equipped with a 12 μm PTFE filter cloth
under a positive pressure of nitrogen (2−5 psig). The mother
liquors were transferred to a 250 L vessel. The 400 L vessel was
rinsed with a solution of IPA (56.0 L, 8.0 vol) and water (14.0
L, 2.0 vol), the cake was washed with this solution, and the
wash solution was collected in a 250 L vessel. The 400L vessel
was rinsed with water (21.0 L, 3.0 vol), the cake was washed
with this water, and the wash solution was collected in a 250 L
vessel. The combined mother liquors and wash solutions were
assayed for AMG 925 HCl losses. The filtration and wash
process took approximately 8 h. The cake was dried under
positive nitrogen pressure (dry nitrogen, 0−5 psig) for 3 days.
The water content (KF) of the solids was measured to be
14.5%. The cake was dried using wet nitrogen (measured at
30−45% relative humidity with a hygrometer) for 14 h. The
water content (KF) was 9.6%, which met in process testing
specifications. The levels of IPA (actual 611 ppm, IPT ≤ 4000
ppm) and acetic acid (actual 3800 ppm, IPT ≤ 4000 ppm) of
the solids also met in process testing specifications, and the
material was packaged. AMG 925 HCl was isolated in 92.5%
yield (7.96 kg), 99.1% overall mass balance, 83.8 wt % AMG
925, 99.75 LCAP AMG 925, 6.2 wt % Cl, 9.6 wt % water, 3800
ppm AcOH, d10 4.0 μm, d50 15.2 μm, d90 38.8 μ, Vm 18.7 μm,
BET surface area 1.5 m2/g. 1H NMR (400 MHz, acetic acid-d4,
mixture of two rotamers at 20 °C) 9.63 (s, 1H), 9.56 (s, 1H),
8.71−8.76 (m, 1H), 8.60−8.66 (m, 1H), 8.20−8.29 (m, 1H),
7.90−7.98 (m, 1H), 4.90−5.01 (m, 1H), 4.86 and 4.70 (2
singlets, 2H), 4.53 and 4.51 (2 singlets, 2H), 4.05 and 3.82 (2
triplets, 2H, J = 6 Hz), 3.11−3.26 (m, 2H), 2.68 (q, 2H, J = 12
Hz), 1.95−2.13 (m, 4H), 1.74 (br s, 1H), 1.36 (q, 2H, J = 12
Hz), 1.06 (d, 3H, J = 8 Hz); 13C NMR (100 MHz, acetic acid-
d4, mixture of two rotamers at 20 °C) 174.9, 174.8, 161.3,
161.2, 160.5, 157.5, 151.6, 151.5, 149.3, 148.9, 145.5, 138.1,
136.0, 129.3, 129.2, 127.1, 126.6, 120.9, 116.7, 116.6, 110.8,
110.7, 63.0, 62.9, 59.3, 46.4, 45.3, 43.2, 41.3, 36.9, 34.3, 32.6,
31.3, 30.6, 24.4; exact mass [C26H29N7O2 + H]+: calculated =
472.2461, measured = 472.2451.
Structural data for ester 15: 1H NMR (400 MHz, acetic acid-
d4, mixture of two rotamers at 20 °C) δ 9.51 (s, 1H), 9.49 (s,
1H), 8.69−8.77 (m, 1H), 8.49−8.55 (m, 1H), 8.41−8.47 (m,
1H), 7.78−7.86 (m, 1H), 4.87−5.02 (m, 3H), 4.83 and 4.74 (2
singlets, 2H), 4.00 and 3.86 (2 triplets, 2H, J = 6 Hz), 3.00−
3.18 (m, 2H), 2.70 (q, 2H, J = 12 Hz), 2.17 (s, 3H), 1.95−2.11
(m, 4H), 1.74 (br s, 1H), 1.36 (q, 2H, J = 12 Hz), 1.07 (d, 3H,
J = 8 Hz); exact mass [C28H31N7O3]+: calculated = 514.2567,
measured = 514.2564.
I
DOI: 10.1021/op500367p
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Communication
(16) The amidation step success was shown to be dependent on the
elimination of these salts from starting material 4.
(17) Methanesulfonic acid was also evaluated to operate the
carbamate cleavage but offered a lower assay yield (<80%).
(18) Lithium hydroxide and ammonium hydroxide solutions were
also tested; however, their use led to the isolation 4 with high levels of
salts.
ASSOCIATED CONTENT
■
S
* Supporting Information
1
Copies of H and 13C NMR spectra, XRPD, DVS plot, and
PSD graph. HPLC method description and retention times of
process intermediates. This material is available free of charge
via the Internet at http://pubs.acs.org.
(19) All processes discussed in this manuscript have undergone a
thorough process safety review prior to being run on kilogram scale.
(20) No exotherm was observed upon heating up to 150 °C.
(21) As much as 8 LCAP of 13 were observed in nondegassed small
scale experiments.
AUTHOR INFORMATION
Corresponding Author
*E-mail: scaille@amgen.com.
■
(22) Patton, D. E.; Drago, R. S. J. Chem. Soc., Perkin Trans, 1 1993,
1611 As discussed in this report, 5-hydroperoxy-1-methylpyrrolidin-2-
one is generated by oxidation of NMP at elevated temperatures in the
presence of oxygen but absent from purchased lots of NMP.
(23) Of note is the fact that CDI only provide AMG 925 in less than
5% assay yield in the absence of catalyst HOBt or Oxyma.
(24) The solubility of diisopropylurea in NMP is greater than 200
mg/mL.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank Will Trieu and David Yeung for help with the GMP
manufacturing operations as well as Shawn D. Walker for
helpful discussions.
(25) Subiros-Funosas, R.; Prohens, R.; Barbas, R.; El-Faham, A.;
Albericio, F. Chem.Eur. J. 2009, 15, 9394−9403.
(26) Wehrstedt, K. D.; Wandrey, P. A.; Heitkamp, D. J. Hazard.
Mater. 2005, 126, 10−12.
(27) Oxyma also has high solubility in NMP (>200 mg/mL) and
consequently is fully rejected upon filtration of the product. One
equivalent of this reagent was used for convenience to continue the
development of this process for this first-in-human program.
(28) The structure of this impurity was assigned by mass
spectroscopy (MW 529 g/mol).
REFERENCES
■
(1) Pratz, K. W.; Levis, M. J. Curr. Drug Targets 2010, 11, 781−789.
(2) Wang, L.; Wang, J.; Blaser, B. W.; Duchemin, A. M.; Kusewitt, D.
F.; Liu, T.; Caligiuri, M. A.; Briesewitz, R. Blood 2007, 110, 2075−
2083.
(3) Li, Z.; Wang, X.; Eksterowicz, J.; Gribble, M. W.; Alba, G. Q.;
Ayres, M.; Carlson, T. J.; Chen, A.; Chen, X.; Cho, R.; Connors, R. V.;
DeGraffenreid, M.; Deignan, J. T.; Duquette, J.; Fan, P.; Fisher, B.; Fu,
J.; Huard, J. N.; Kaizerman, J.; Keegan, K. S.; Li, C.; Li, K.; Li, Y.;
Liang, L.; Liu, W.; Lively, S. E.; Lo, M.-C.; Ma, J.; McMinn, D. L.;
Mihalic, J. T.; Modi, K.; Ngo, R.; Pattabiraman, K.; Piper, D. E.;
Queva, C.; Ragains, M. L.; Suchomel, J.; Thibault, S.; Walker, N.;
Wang, X.; Wang, Z.; Wanska, M.; When, P. M.; Weidner, M. F.;
Zhang, A. J.; Zhao, X.; Kamb, A.; Wickramasinghe, D.; Dai, K.;
McGee, L. R.; Medina, J. C. J. Med. Chem. 2014, 57, 3430−3449.
(4) (a) Hartwig, J. F. Acc. Chem. Res. 2008, 41, 1534. (b) Surry, D. S.;
Buchwald, S. L. Angew. Chem., Int. Ed. 2008, 47, 6338. (c) Busacca, C.
A.; Fandrick, D. R.; Song, J. J.; Senanayake, C. H. Adv. Synth. Catal.
2011, 353, 1825.
(29) Performing the amidation at 60 °C provided a 65 °C buffer
relative to the reaction mixture exothermic onset (125 °C, 45 J′g
energy released).
(30) The AMG 925 supernatant concentration upon filtration was
∼2 mg/mL.
(31) FBRM data was gathered and corroborated the minimization of
fine particles using heat-cycles.
(32) Subsurface N2 sparge.
(5) Purity by liquid chromatography area percent.
̀ ́
(6) Awad, H.; Mongin, F.; Trecourt, F.; Queguiner, G.; Marsais, F.;
Blanco, F.; Abarca, B.; Ballesteros, R. Tetrahedron Lett. 2004, 45,
6697−6701.
(7) For a related example of Negishi cross-coupling performed on
kilogram scale, see Denni-Dischert, D.; Marterer, W.; Banziger, M.;
Yusuff, N.; Batt, D.; Ramsey, T.; Geng, P.; Michael, W.; Wang, R.-M.
B.; Taplin, F.; Versace, R.; Cesarz, D.; Perez, L. B. Org. Process. Res.
Dev. 2006, 10, 70−77.
(8) Yield corrected for isolated product wt %.
(9) In such a case, no more than 10 L of solvent per kilogram of
intermediate have to be used in order to maintain a solution during
processing.
(10) Starting materials 1 and 2 were treated with various bases
(KHMDS, NaHMDS, LiHMDS, KOtBu, NaOtBu, LiOtBu) in NMP
(10 volumes) at 100 °C.
(11) Fors, B. P.; Watson, D. A.; Biscoe, M. R.; Buchwald, S. L. J. Am.
Chem. Soc. 2008, 130, 13552−13554.
(12) The specification for palladium in the drug substance (AMG
925 hydrochloride) was set to less than 100 ppm for clinical use
considering that the permitted daily palladium exposure is 100 ug/day
(Guidelines for Elemental Impurities, Q3D, FDA, July 26, 2013) based
on chronic use, and the maximum projected QD dose for Ph1 for this
material was 1 g/day.
(13) Garrett, C. E.; Prasad, K. Adv. Synth. Catal 2004, 346, 889.
(14) Commercialized as SiliaMetS Thiourea by Silicycle.
(15) The use of 15 volumes of solvent ensured that compound 3
stayed in solution during filtration of the silicyle−thiourea as the
temperature of the mixture upon filtration decreased to >30 °C.
J
DOI: 10.1021/op500367p
Org. Process Res. Dev. XXXX, XXX, XXX−XXX