Development of a scalable synthesis of mevidalen (LY3154207), an orally available positive allosteric modulator of the human dopamine D1 receptor

Article abstract of DOI:10.1021/acs.oprd.0c00229

The evolution from early medicinal chemistry to large-scale production of the chemical synthesis of Lilly D1-positive allosteric modulator (PAM) mevidalen (LY3154207) and its hydroxybenzoate co-crystal is outlined. The issues and steps taken to resolve them are outlined across several generations of synthesis, including unexpected issues that arose during cryogenic addition of MeLi to a key imine intermediate and the use of flow chemistry to enable large-scale production. Ultimately, a process that was used to deliver >100 kg of API to support ongoing clinical trials is described.

Full text of DOI:10.1021/acs.oprd.0c00229

Subscriber access provided by UNIV OF NEW ENGLAND ARMIDALE
Full Paper
Development of a Scalable Synthesis of Mevidalen (LY3154207), an Orally
Available Positive Allosteric Modulator of the Human Dopamine D1 Receptor
Jeffery Richardson, Kevin Paul Cole, Amy C. DeBaillie, Junliang Hao,
Jared Piper, Peter Houghton, John R Rizzo, Mo Jia, Peng Liu, Ping Huang,
Katarina Kadlecikova, Conchita Fernandez Garcia, and Daniel S Richards
Org. Process Res. Dev., Just Accepted Manuscript • DOI: 10.1021/acs.oprd.0c00229 • Publication Date (Web): 22 Sep 2020
Downloaded from pubs.acs.org on September 22, 2020
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 38
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
Development of a Scalable Synthesis of Mevidalen (LY3154207), an
Orally Available Positive Allosteric Modulator of the Human Dopamine
D1 Receptor
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
c
Jeffery Richardson a*†1, Kevin P. Cole b, Amy C. DeBaillie b, Junliang Hao , Jared Piper†3, Peter
Houghton b, John Rizzo b, Mo Jia d, Peng Liu d, Ping Huang d, Katarina Kadlecikova†4 e, Conchita
F. Garcia e and Daniel S. Richards f†2
.
a Eli Lilly and Company, Erl Wood Manor, Sunninghill Road, Windlesham, Surrey, GU20 6PH,
United Kingdom.
b Small Molecule Design and Development, Lilly Research Laboratories, Eli Lilly and Company,
Lilly Corporate Center, Indianapolis, Indiana 46285, United States.
c Discovery Chemistry Research and Technologies, Eli Lilly and Company, Indianapolis, Indiana
46285, United States.
d
STA Pharmaceutical, No. 1 Building, #288 Fu Te Zhong Lu, Wai Gao Qiao Free Trade Zone,
Shanghai, P. R. China.
e
Almac Sciences, Almac House, 20 Seagoe Industrial Estate, Craigavon, BT63 5QD, United
Kingdom.
f
AMRI UK Ltd, Erl Wood Manor, Sunninghill Road, Windlesham, Surrey, GU20 6PH, United
Kingdom.
ACS Paragon Plus Environment
1

Organic Process Research & Development
Page 2 of 38
1
2
3
4
5
6
7
8
OH
mevidalen
LY3154207
>100 kg produced
Pictet-Spengler,
Cryogenic Flow
MeLi Addition
Br
Br
9
CO2H
NH2
OPG
OH Cl
N
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
NH
O
Cl
Mevidalen Process Evolution: Medicinal Chemistry
Through three GMP Production Campaigns!
TOC Graphic
ACS Paragon Plus Environment
2

Page 3 of 38
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
ABSTRACT The evolution from early medicinal chemistry to large-scale production of the
chemical synthesis of Lilly D1 positive allosteric modulator (PAM) mevidalen (LY3154207) and
its hydroxybenzoate cocrystal is outlined. The issues and steps taken to resolve them are outlined
across several generations of synthesis, including unexpected issues that arose during cryogenic
addition of MeLi to a key imine intermediate and the use of flow chemistry to enable the large-
scale production. Ultimately a process that was used to deliver >100 kg of API to support ongoing
clinical trials is described.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
KEYWORDS: continuous processing, flow chemistry, tetrahydroisoquinoline, mevidalen, process
development, co-crystal, D1 PAM, LY3154207.
1. INTRODUCTION
Activation of the D1 receptor holds significant promise for the treatment of several neurological
disorders. The positive allosteric modulation approach to increase D1 receptor activity is especially
promising since it has the potential to address several issues associated with known D1 receptor
agonists. Binding at an allosteric site of the D1 receptor potentially addresses pharmacological
selectivity issues associated with orthosteric agonists, and also provides an opportunity to address
the chemical stability issues of some known D1 agonists by requiring a different pharmacophore.
Additionally, the more physiologically relevant mode of action of a D1 PAM may result in a lower
propensity for overstimulation and tolerance development, as well as a potentially better safety
profile.1 For those reasons the D1 positive allosteric modulator mevidalen (LY3154207, 3) and
mevidalen hydroxybenzoate (3•HBA) is currently in phase 2 development by Eli Lilly and
Company for the treatment of Lewy body dementia (PRESENCE, NCT03305809).2
As part of the journey of this molecule from early discovery chemistry through preclinical
toxicology and ultimately into patients, significant investment was required to enable the supply
of material at sufficient scale meeting the relevant quality standards. This article highlights the
ACS Paragon Plus Environment
3

Organic Process Research & Development
Page 4 of 38
1
2
3
4
5
6
evolution of our synthetic chemistry strategy for this molecule from early discovery to phase 2
clinical testing.
7
8
2. DISCUSSION AND RESULTS
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In the first part of this account, focus will be on the synthesis of intermediate 2 (Scheme 1). The
second part will discuss the conversion of 2 to 3 and its 4-hydroxybenzoic acid (HBA) co-crystal
(3•HBA).
2.1. Discovery Chemistry Synthesis of 2.
Our overall strategy for the synthesis of 3 involved two main phases: the synthesis of an
intermediate that already had key structural features installed and would allow access to a variety
of potential structure-activity relationship (SAR) analogues1 and the conversion of this
intermediate to 3 (Scheme 1). Intermediate 2 was identified as being capable of achieving both
goals.
OH
Br
Br
O
OPG
OH Cl
OH
NH
N
NH2
O
Cl
1
2
mevidalen
LY3154207
3
Scheme 1: Overall synthetic strategy for mevidalen (LY3154207, 3).
Early medicinal chemistry routes to analogous compounds that preceded 3 in our SAR relied on
deprotonation of a Boc protected tetrahydroisoquinoline (THIQ) with LDA and subsequent
trapping with MeI to install the C-1 methyl group. This approach was unattractive for large scale
work for several reasons, including the cryogenic conditions employed, poor diastereoselectivity
and difficult isomer separation. When attempted on relevant substrate 4 (Scheme 2), a mixture of
diastereomers and unconsumed starting material was obtained. This mixture could only be
separated by deprotection and preparative HPLC affording the desired diastereomer 5 in 28% yield
over the 2 steps.
ACS Paragon Plus Environment
4

Page 5 of 38
Organic Process Research & Development
1
2
3
4
5
6
Step 1
Br
Br
1) LDA (1.5 eq), -78 oC
2) MeI (2 eq), -78 oC
Br
OH
Boc
OH
OTBS
+
N
N
N
7
Boc
Step 2
Boc
8
9
1) TBAF, THF, rt
2) prep HPLC
6
5
4
28%
11%
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 2: Early approach to installation of C-1 methyl group.
Efficient installation of the C-1 methyl group was one of the key route issues that needed to be
addressed and early work focused on alternate methods. Ultimately, the benzylic oxidation to form
imine 16, followed by addition of MeMgCl at low temperature allowed the highly
diastereoselective installation of the C-1 Me group (Scheme 3).3
Br
O
Br
O
Br
O
MeOCOCl,
AcCl, MeOH, rt
100%
CH2Cl2, H2O, NaHCO3
O
O
OH
O
HN
99%
·
NH2
NH2
1
8
O
7
Br
O
paraformaldehyde,
AcOH,
H2SO4, 35 °C
Br
O
O
Br
O
MeOH, AcCl,
0 oC to rt
O
HCl (5N aq), reflux
85%
OH
O
OH
•HCl
NH
100%
•HCl
N
NH
91%
9
10•HCl
11•HCl
LiBEt3H (12 eq),
Br
O
Br
CF3CO2H (66 eq).
CH2Cl2
Br
THF, 0 oC to rt
Boc2O, 1,4-dioxane,
NaHCO3, rt
O
OH
OH
71%
98%
N
NH
N
Boc
•CF3CO2H
Boc
14•TFA
12
13
1) NCS, Et2O
2) concentrate
3) KOH, MeOH
Br
Br
Br
MeMgCl (8.4 eq), THF,
Et2O, -78 °C to rt
TBSCl, 4-DMAP,
imidazole, CH2Cl2
OTBS
OTBS
OTBS
NH
82%
77%
33% over 2 steps
NH
N
>9:1 dr
15
16
17
Scheme 3: Initial discovery chemistry route to intermediate 17.
Synthesis of 17 began with (R)-2-bromophenylalanine (1). Esterification followed by methyl
carbamate formation afforded the fully protected 8, and the THIQ ring was subsequently formed
through a Pictet−Spengler reaction4 (Scheme 3). Over several steps, 9 was converted to imine 16
after N-chlorination and base-mediated dehydrohalogenation.5 Finally, addition of MeMgCl in
diethyl ether to this imine delivered 17 with good trans diastereoselectivity (>9:1 dr).3
ACS Paragon Plus Environment
5

Organic Process Research & Development
Page 6 of 38
1
2
3
4
5
6
7
8
9
This intermediate was prepared in 11 overall steps and 11% overall yield. While many of the steps
were high yielding and gram quantities were delivered for early SAR studies, there were many
arguably unproductive protecting group manipulations with potential opportunities to shorten the
sequence and eliminate numerous chromatographic separations. Additionally, amine 17 is an oil
which would present handling, isolation and potential stability challenges at this key hold point in
the synthesis.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
2.2 Preclinical toxicology route to 2.
In order to prepare 3 (LY3154207) in quantities sufficient for preclinical toxicology studies (ca.
170 g), the initial focus was on reducing the number of chromatographic purifications, streamlining
the protecting group manipulations and installing the C-1 Me group as part of the cyclisation
strategy towards this key intermediate. While one of the key issues had been resolved, and the
methyl group installation afforded good trans-selectivity and retained the ee of the starting amino
acid throughout, the synthesis was longer than necessary with too many protecting group
manipulations.
Only minor changes were made to the first two steps and early development activities focused on
attempts to install the methyl group as part of the skeletal construction using Bischler-Napieralski
or Pictet-Spengler approaches (Scheme 4). If this could be achieved, there was potential to
eliminate both the early carbamate protection step and the 2-step methyl group introduction that
occurred at the end of this sequence.
Br
Br
Br
POCl3, etc.
R1 = Ac
MeCHO
acid catalyst
R2
NR1
R2
R2
NHR1
N
Pictet-Spengler
Bischler-
Napieralski
Scheme 4: Strategies for the introduction of the C-1 methyl group during cyclisation.
Many attempts were made to perform the Pictet-Spengler cyclisation using acetaldehyde in place
of formaldehyde (R2 = CO2Me, R1 = H, Boc, CO2Me) but in no case was the desired cyclisation
product formed, even in small quantities.
ACS Paragon Plus Environment
6

Page 7 of 38
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
Attempts to form the imine directly through a Bischler-Napieralski type cyclisation were equally
unsuccessful under numerous conditions although the cyclisation as reported by Larsen and co-
workers6 performed reasonably well on gram scale using phenylalanine methyl ester as a model
substrate. No further action was taken with this potentially promising route, as we felt that
significant additional optimization would be required to make it viable. Instead, we focused on
streamlining the synthesis through improved selection of protecting groups. These studies revealed
that the protecting group on nitrogen was extremely important for the Pictet-Spengler reaction as
it needed to be stable under the acidic conditions; after numerous attempts with other groups, it
was clear that the methyl carbamate that was originally employed in the first-generation medicinal
chemistry synthesis (Scheme 3) was the best in terms of yield and acid stability.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
During the Pictet-Spengler reaction partial hydrolysis of the methyl ester occurs, but the mixture
of the acid and ester can be telescoped into the carbamate deprotection (Scheme 5). Treating this
mixture with aq. HCl serves to remove the carbamate and completes the ester hydrolysis allowing
the hydrochloride salt (10) to be isolated by filtration in 89% yield.
Reduction of the carboxylic acid was achieved using lithium aluminium hydride (LAH) and the
crude amino alcohol 14 can be telescoped into the TBS (tert-butyl dimethylsilyl) protection.
Oxidation to imine 16 was achieved in THF rather than diethyl ether, which was originally used
and was not desirable for work at scale, and subsequent addition of Grignard reagent MeMgCl
afforded 17 with good levels of diastereoselectivity. Despite significant effort, including
purification of 16 via silica gel chromatography, it was not possible to reduce the amount of
Grignard used in this final step below 8.4 equivalents.
ACS Paragon Plus Environment
7

Organic Process Research & Development
Page 8 of 38
1
2
3
4
5
6
Br
O
O
Br
O
Br
O
MeOCOCl, CH2Cl2,
H2O, NaHCO3
paraformaldehyde,
AcCl, MeOH, 65 °C
96%
AcOH, H2SO4, 35 °C
O
O
OH
O
HN
87%
80%
·
NH2
NH
2 HCl
7
1
7•HCl
8
8
9
Br
O
Br
O
Br
O
Br
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
HCl (aq), 95 °C
89%
LiAlH4, THF, -60 °C
81%
O
O
OH
+
OH
OH
NH
N
N
O
NH
HCl
O
O
18
9
10•HCl
14
1) NCS, THF
2) concentrate
3) KOH, MeOH
Br
MeMgCl (8.4 eq) in THF,
Et2O, -65 °C
Br
Br
TBSCl, 4-DMAP,
imidazole, CH2Cl2, DMF
OTBS
OTBS
OTBS
NH
96%
73%
81%
NH
N
15
16
17
Scheme 5: Second generation discovery chemistry route to 17.
This streamlined route afforded 17 in 27% overall yield and 8 steps compared to the first-
generation synthesis (11 steps, 11% overall yield) with all nine chromatographic purifications
removed over the course of the route through telescoping or isolation of products as solids. The
final step required chromatographic purification primarily because 17 is an oil and all efforts to
purify it by other means were unsuccessful. This route was used to deliver 3 kg of 17 at one of our
large-scale partners.7 The fact that 17 was an oil presented a number of storage, handling and
purification challenges which we sought to obviate through preparation of a stable salt.
Hydrochloric acid was the only acid examined at this stage and a study of solvents showed that in
most solvents, a significant amount of silyl deprotection occurred, but by using iPrOAc or PhMe
as solvent, the hydrochloride salt (17•HCl) could be isolated in high yield with no desilylation
product (19•HCl) observed (Table 1). Recognizing that exposure of the slurry to atmospheric
moisture during the salt formation could potentially cause desilylation to occur during larger scale
processing, the vials were then left open to atmosphere overnight; in a few solvents significant
deprotection began to occur. As a final attempt to assess potential risk, a small amount of water
was added to each salt formation. This caused significant dissolution of the solids, but in PhMe
and iPrOAc essentially no deprotection was observed. The solid isolated was not appreciably
hygroscopic once collected by filtration and this served as a suitable way to store this intermediate
during the discovery campaigns.
Table 1. Formation of 17•HCl.a
ACS Paragon Plus Environment
8

Page 9 of 38
Organic Process Research & Development
1
2
3
4
5
6
Br
Br
Br
4N HCl in dioxane (1 eq)
solvent (10 vol), rt
OTBS
OH
OTBS
+
NH
NH
NH
7
.HCl
.HCl
8
9
17•HCl
19•HCl
17
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Entry
Solvent
Ratio 17:19 under
Ratio 17:19
open to air c
100:0
Ratio 17:19
b
N2
(+ 1 drop water) b
100:0
1
2
3
3
5
6
iPrOAc
PhMe
100:0
100:0
100:0
100:0
100:0
100:0
100:0
100:0
iPrOH
44:56
28:72
MTBE
2-MeTHF
MeCN
100:0
85:14
38:63
43:57
6:94
6:94
a Ratios determined by HPLC analysis. b obtained after 1 h. c obtained after 19 h.
Head-to-head comparison showed that the iPrOAc crystallization conditions were able to purge
some impurities contained in the freebase, allowing a batch of 17 with 94.8% purity by HPLC to
be upgraded to >98% purity during the salt formation. These conditions were selected for further
use.
2.3. Process Development Synthesis of 2.
As mevidalen entered clinical development, a route that could provide >100 kg of 17, or an
equivalent intermediate, was required. While alternate routes would be sought, it was felt that the
discovery route (Scheme 5) was suitable for this stage of development since the unnecessary
protecting group manipulations had been removed from the route. In order to prepare the synthesis
route for production scale, attention was turned to improvement of the individual steps in the
sequence. Since entering clinical development, three discrete manufacturing campaigns using this
technology have taken place over several years to fund our clinical studies. The following section
describes the development chemistry and the results from these campaigns.
2.3.1 Steps 14 (Synthesis of 10•HCl)
ACS Paragon Plus Environment
9

Organic Process Research & Development
Page 10 of 38
1
2
3
4
5
6
7
8
9
The production route did not significantly differ from the previous route over the first four steps,
only minor changes to stoichiometry and workups were implemented to afford hydrochloride
10•HCl. These steps were robust and tolerated residual solvents from previous processing steps,
enabling the sequence to be fully telescoped.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
While esterification of 1 had performed well in the first manufacturing campaign using the
conditions described in the discovery route, the long reaction time (typically ~20 h) and high
MeOH volumes prompted additional optimization work. It was found that by replacing acetyl
chloride with thionyl chloride (Scheme 6), the reaction could achieve complete conversion in 3 h
at laboratory scale and could be performed at higher concentration (4 V; V = liters of solvent per
kg of limiting reagent) with no stability issues over 44 h. In the methyl carbamate formation (step
2), a biphasic mixture of CH2Cl2 and aq. NaHCO3 was used in first two campaigns, which required
20 V CH2Cl2 and 12 V water.8 Attempts to reduce these volumes lead to lower purity 8. It was
assumed that the low aqueous solubility of NaHCO3 drove the use of large volume of solvent and
so various bases were then screened to replace NaHCO3. K2CO3 showed comparable results and
allowed the overall solvent volume to be reduced from 32 V to 8 V in the subsequent campaign.
No further changes were made in the 3rd campaign when applying this four-step sequence.
Optimization of the Pictet-Spengler reaction centered around practical improvements and ensuring
robustness. Initially the workup used 40 V of solvent, so efforts were made to lower the reaction
solvent volume to enable waste reduction in the workup. Indeed, it was possible to reduce the
volume from 7.5 V AcOH and 2.5 V sulfuric acid to 3 V and 1.5 V, respectively. This modification
worked well, although the yield of one of the production runs in campaign 3 was reduced due to
difficulty in the aqueous separation (72% vs average yield of 82% for other batches in campaign
3).
Little change was made to the acidic hydrolysis to form 10•HCl (Step 4), but several changes were
made to the workup and isolation. Stress testing showed both acetic acid and sulfuric acid could
be carried through from the previous step with no detrimental effects. The previous route (section
2.2) involved evaporation to dryness after extraction to afford a sticky solid and, in turn, a
heterogeneous reaction mixture. It was found that the extract from the Pictet-Spengler reaction
could be taken directly into this hydrolysis thus alleviating the issues of handling the sticky solid,
which would have proven challenging on scale.
ACS Paragon Plus Environment
10

Page 11 of 38
Organic Process Research & Development
1
2
3
4
5
6
Campaign 1&2
CH2Cl2, H2O, NaHCO3,
methyl chloroformate
Br
O
O
paraformaldehyde,
AcOH, H2SO4, 25 °C
Br
O
Br
O
Campaign 1
AcCl, MeOH, 65 °C
O
O
OH
O
HN
Campaign 2
SOCl2, MeOH, 65 °C
Campaign 3
CH2Cl2, H2O, K2CO3,
methyl chloroformate
·
NH2
NH
2 HCl
1
7
7•HCl
8
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Br
O
Br
O
O
Br
O
1) HCl (aq), 100 °C
2) filter
Campaign 1
Campaign 2
Campaign 3
O
O
OH
+
OH
HCl
4 Batches 65 Kg each
80-85% (4 Steps)
259 Kg produced
>99% purity
2 Batches 68 & 70 Kg
81-87% (4 Steps)
142 Kg produced
>99% purity
4 Batches 75-140 Kg
72-84% (4 Steps)
431 Kg produced
98.5-99% purity
98-99% ee
N
N
O
·
NH
10•HCl
O
18
9
98% ee
98% ee
Scheme 6: Production process for 10•HCl.
At the end of these four telescoped steps, the material was obtained by filtration in 97–98% purity
as determined by HPLC, and it was found that this could be increased to >99.0% HPLC area by
slurrying the solids in CH2Cl2.
2.3.2 Steps 57 (Synthesis of 21•oxalate)
The reduction of carboxylic acid 10•HCl to alcohol 14 was achieved at laboratory scale using
LAH but this has considerable drawbacks on larger scale.9 We sought to eliminate the use of LAH
and also improve the workup to avoid issues associated with the aluminium residues that are
generated. A survey of hydride-based reductants that were capable of the direct transformation of
10 to 14 was performed and Red-Al® was found to be a good replacement for LAH (Scheme 7),
with borane complexes generally affording lower purity profiles by HPLC. Freebasing the starting
material added several additional processing steps and did not improve the reaction profile and so
was deemed unnecessary, even though the additional acid would quench the reductant. The
corresponding methyl ester (11) reduced cleanly with Red-Al®, but it proved difficult to control
unwanted hydrolysis of this ester, hence it was decided to proceed via direct reduction of 10•HCl.
One of the major issues with performing an aluminium hydride reaction on scale is the quench and
subsequent workup. In our discovery route, a Fieser workup was employed allowing the
aluminium residues to be removed by filtration.10 It was found that simply quenching the reduction
with 10 V of saturated aq. NaCl gave good layer separation with minimal product loss to the
aqueous layer. Addition of n-heptane anti-solvent then caused precipitation of 14 from the reaction
mixture, thus significantly simplifying the workup and isolation. The modified process using Red-
Al® was used in campaign 1 to prepare 134 kg of 14.
ACS Paragon Plus Environment
11

Organic Process Research & Development
Page 12 of 38
1
2
3
4
5
6
7
8
9
Despite Red-Al® having been successfully applied to a large-scale preparation of 14, a significant
impurity that arose was the debrominated side product. Pilot reactions suggested that this could be
avoided by using sodium borohydride for reduction of the methyl ester (11), but it was not possible
to avoid partial ester hydrolysis during the Pictet-Spengler reaction. Therefore, we chose to
implement a re-esterification step in later campaigns, with the benefits gained in the reduction
outweighing the disadvantages of an additional step (Scheme 7).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The conditions that were used to prepare 7•HCl from 1 (Scheme 6) were equally applicable to the
formation of 11•HCl from 10•HCl and successfully applied in campaign 2 (Scheme 7). It was
found that 11, obtained after free basing in CH2Cl2/aqueous K2CO3 could be solvent swapped into
methanol and reduced cleanly with portion wise additions of NaBH4. Thermal hazard analysis of
the borohydride reduction revealed a high adiabatic temperature rise (89 °C) and a maximum heat
flow of 455 W/kg. Portion wise addition of NaBH4 was used to help enable the safe venting of
hydrogen gas generated during the addition. This new process was used to deliver 161 kg of
21•oxalate in our second production campaign.
Br
Br
Br
O
Br
Et3N, NCS, CH2Cl2
then Oxalic acid, EtOAc
TBDPSCl, imid.
CH2Cl2,20-30 o
Red-Al,
THF, 7-20 °C
C
OTBDPS
OTBDPS
O
OH
HCl
OH
Campaign 1
4 batches 60-70 kg ea.
134 kg produced
59-69%
·
NH
20
N
NH
10.HCl
NH
O
14
OH
100% ee
OH
21•oxalate
Campaign 2
Campaign 1
Campaign 2
1 Batch 84 Kg
70% (from 11)
161 Kg produced
93% purity
K2CO3, CH2Cl2/H2O
then NaBH4, MeOH
2 batches 142 Kg
84 Kg produced
Campaign 2
AcCl, MeOH,
65 °C
2 Batches 65 & 69 Kg
68-75% (from 11)
253 Kg produced
95-96% purity
Br
O
OMe
HCl
79-82% (2 Steps)
>99% ee
>99% ee
NH
·
11•HCl
Scheme 7: Production process for 21•oxalate.
The protection of the primary alcohol, as described in the discovery synthesis, posed a few issues;
firstly, DMF was used as a co-solvent in the protection, leading to issues in its removal at scale.
Secondly, as described above (section 2.2), the TBS group was prone to spontaneous deprotection
under certain conditions, which suggested that it might be advantageous to find a more stable
alternative. Finally, we reasoned that diastereoselectivity could be potentially improved in the
organometallic addition to imine 16 if a more sterically encumbering protecting group were used.
The bulkier and more stable tert-butyldiphenylsilyl (TBDPS) group was considered and so a small
batch was carried out to test the organometallic addition to imine 21. While the TBPDS group
ACS Paragon Plus Environment
12

Page 13 of 38
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
fulfilled the majority of these criteria, it was not clear whether it had a positive impact on
diastereoselectivity in comparison to factors such as solvent and additive, however, the advantages
outlined above were sufficient and so TBDPS was selected in place of TBS. For the protection
step, a small solvent screen showed that THF gave a heterogeneous reaction mixture and lower
assay purity while CH2Cl2 and DMF gave homogeneous mixtures and higher assay purity. As a
result, CH2Cl2 was selected as this was a suitable solvent for telescoping into the subsequent
oxidation reaction. Attempts to replace imidazole with Et3N or iPr2NEt met with little success,
affording slower reactions and low in situ purity. Initial studies showed that this mixture could be
directly telescoped into the oxidation step and that an aqueous workup between these steps did not
improve the outcome and was therefore unnecessary. In the largest scale discovery synthesis
(Scheme 5), THF was used as solvent, with KOH/MeOH as base/cosolvent, but the reaction also
performed well in CH2Cl2 allowing the telescoping of the protection directly into the oxidation.
The oxidation occurs in two discrete stages: chlorination followed by elimination to the imine. By
adding triethylamine and cooling to 0 °C before addition of NCS, a significant improvement in
assay purity was achieved (92.0–95.6% compared to the original procedure, which gave 81.6%
purity).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Aqueous workup served only to remove salts and excess base. Little improvement to the impurity
profile was realized. We therefore sought a strategy to isolate the imine that would provide good
impurity control going into the final step of the sequence. Initial studies showed that the oxalic
acid and HCl salts formed easily, but that the wet cake of the oxalate salt was more stable and was
therefore selected for additional study. This salt (21•oxalate) was highly soluble in CH2Cl2 and
THF, but EtOAc gave suitable solubility for crystallization and excellent impurity rejection. The
major impurities that were rejected during this crystallization were the isoquinoline that was
formed by overoxidation of 21 and the dehalogenated imine, which was believed to be formed
during the Red-Al® reduction and is further controlled in the isolation of 22.
2.3.3 Steps 89 (Synthesis of 22)
With a stable imine salt in hand, where the input purity can be highly controlled, addition of
organometallic reagents was studied. One of the major issues with the original procedure was that
a significant excess of Grignard reagent was required to achieve full conversion and the reaction
was quite slow, taking >16 h to reach completion. Suspecting that the imine was not especially
ACS Paragon Plus Environment
13

Organic Process Research & Development
Page 14 of 38
1
2
3
4
5
6
7
8
9
electrophilic,11 a survey of potential activators was performed. From this study it was clear that
many of the activators studied did not accelerate the reaction, however the use of
chlorotrimethylsliane (TMSCl) or BF3•OEt2 accelerated the reaction but resulted in reduced
diastereoselectivity (3:1), presumably through formation of an iminium ion intermediate.12,13
Increasing the reaction temperature also accelerated the reaction, but significantly decreased the
assay purity of the product. Given the relatively clean reaction profile, (TMSCl) was selected for
further study (Table 2) and it was found that by lowering the temperature to -60 °C improves the
diastereoselectivity (Entry 2) and this can be further increased by using methyllithium as the
organometallic reagent (Entry 3).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 2: Optimization of organometallic addition to 21.
Br
R-M, Me3SiCl,
THF
Br
OTBDPS
OTBDPS
NH
N
21
22
Entry
R-M
TMSCl
Solvent Temp
22a
21 a
dr a
(equiv.) volume
1
MeMgBr (5 equiv.) 1.1
MeMgBr (5 equiv.) 1.1
20 V
20 V
20 V
20 V
20 V
20 V
20 V
20 V
20 V
5 V
-10 °C 70% 17% 3:1
2
-60 °C 59% 4%
-60 °C 57% 7%
-60 °C 75% 5%
5:1
3
MeLi (5 equiv.)
MeLi (5 equiv.)
MeLi (5 equiv.)
MeLi (5 equiv.)
MeLi (4 equiv.)
MeLi (3 equiv.)
MeLi (2 equiv.)
MeLi (2 equiv.)
MeLi (2 equiv.)
1.1
1.5
0.5
2.0
1.5
1.5
1.5
1.5
1.5
12:1
10:1
4
5
-60 °C 25% 10% n.d.
6
-60 °C 68% 6%
-60 °C 78% 3%
-60 °C 80% 3%
-60 °C 85% 1%
6.9:1
7
12:1
9.2:1
10:1
8
9
10
11
-60 °C 57% 1.3% 8.9:1
-60 °C 68% 3% 7.1:1
10 V
ACS Paragon Plus Environment
14

Page 15 of 38
Organic Process Research & Development
1
2
3
4
12
MeLi (2 equiv.)
1.5
15 V
-60 °C 65% 3%
8.2:1
5
6
a) Determined by HPLC analysis.
7
8
9
The best conditions from this study were those in entry 9, but it was preferred to perform the
reaction at higher concentration for efficient use of the available reactors, and so the conditions
employing 10 V of solvent were selected (entry 11) and it was found that the purity profile could
be increased by further decreasing the temperature of the reaction. Additionally, the oxalate salt
could not be employed directly in the organometallic addition but must be free based with aqueous
potassium carbonate in CH2Cl2 prior to the organometallic reaction with water content below 0.1%
being acceptable for forward processing.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
As with compound 17 (section 2.2), it was desirable to prepare 22 as a salt that could be readily
isolated with additional opportunity to improve the purity of the material, especially diastereomeric
purity. The HCl and oxalic acid salts were examined first since the former had successfully been
used at this stage in the discovery synthesis and the latter formed a salt with imine 21. Using a
sample of 22 that contained significant amount of the undesired diastereomer (3:1 dr) it was found
that in EtOAc, HCl formed a salt which improved purity, but did not change the dr, while oxalic
acid did not form a precipitate. As such, a screen of 17 acids and 5 solvents was performed. Only
a small number of combinations formed solids (Table 3), but the combination of EtOAc and
sulfuric acid afforded a significant improvement in the diastereomer ratio. On further analysis of
the salt formed from these conditions, it became clear that a hemisulfate salt had formed
(22•Hemisulfate).
Table 3. Selected acids screened in the formation of salts (22•X)
Br
Br
HX (1.1 equiv)
Solvent (10 vol)
OTBDPS
•HX
OTBDPS
NH
NH
22
dr 3:1
22•HX
Acid
D-(+)-malic acid
11.6:1
Solvent
c
D-(+)-DBTA
H2SO4
a
a
2-propanol
EtOAc
-
-
a
4:1
-
37:1
ACS Paragon Plus Environment
15

Organic Process Research & Development
Page 16 of 38
1
2
3
4
5
6
7
8
a
a
a
a
Acetone
MeCN
PhMe
-
-
-
-
-
10:1
a
6.1:1
8.5:1b
-
a
9
Diastereomeric ratios of isolated salts, determined by HPLC. a) No solid formed b) 5 V of solvent
used c) 0.55 equiv.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
To robustly remove impurities and undesired diastereomer we sought to recrystallize the
hemisulfate salt. Analysis of a small number of solvents showed that THF/EtOAc was a suitable
solvent system for the crystallization to improve purity and dr, allowing a pilot batch with 93%
assay purity and 15.6:1 dr to be upgraded to 96% assay purity and 117:1 dr.
This sequence was then applied in the production of 102 kg of 22•hemisulfate (Scheme 8).
Br
Br
Br
1) CH2Cl2 ,K2CO3 basification,
solvent exchange to THF
H2SO4, EtOAc
OTBDPS
OTBDPS
OTBDPS
NH
NH
2) MeLi,TMSCl, THF, -65 °C
Aqueous w-up,
Solvent exchange to EtOAc
•1/2 H2SO4
N
HO
O
22•Hemisulfate
Campaign 1
HO
21•Oxalate
O
22
4 Batches 20-80 Kg
35-57% (2 steps)
102 Kg produced
96-98% purity
>99:1 dr
Scheme 8: First production campaign for 22•hemisulfate.
The first clinical supply manufacturing campaign utilized batch mode cryogenic conditions for the
addition of MeLi to imine 21. While a 20 kg batch performed as expected (Table 4, Entry 1),
scaling the addition reaction to 80 kg lot size resulted in a substantial increase in impurities reaction
profile and isolated yield (Entry 2). In order to complete the first campaign to deliver
22•hemisulfate, the batch size was limited to 40 kg and material was processed accordingly
(Entries 36).
Given the unexpectedly poor performance of the MeLi at 80 kg scale, Grignard reagent addition
was revisited in anticipation of the second clinical supply manufacture. It was shown that MeMgBr
could be successfully utilized in batch mode under non-cryogenic conditions to afford acceptable
results (Entry 7). However, superior results were obtained by conversion of the batch MeLi
process to flow mode (Entry 8).
Table 4: Performance of C-1 methyl installation across campaigns 1 & 2.
ACS Paragon Plus Environment
16

Page 17 of 38
Organic Process Research & Development
1
2
3
4
5
6
Br
Br
Br
1) CH2Cl2, K2CO3 basification,
solvent exchange to THF
H2SO4, EtOAc
OTBDPS
OTBDPS
OTBDPS
NH
NH
2) Nucleophille, additive,
solvent, temp
•1/2 H2SO4
N
HO
O
7
8
9
22•Hemisulfate
22
HO
21•Oxalate
O
10
11
12
13
14
15
16
17
Entry
Campaign Input
(kg)
Nu
Mode Solvent
Additive Temp
dra of yield
22
dr a
22•
Hemisulfate
99.7:0.3
99.8:0.2
99.8:0.2
1
2
3
4
5
6
7
8
1
20
MeLi
MeLi
MeLi
MeLi
MeLi
MeLi
MeMgBr
MeLi
Batch THF/MeTHF
TMSCl
TMSCl
TMSCl
TMSCl
TMSCl
TMSCl
none
-75 °C 89:11
-75 °C 87:13
-75 °C 87:13
-75 °C 88:12
-75 °C 87:13
-75 °C 87:13
58%
35%
54%
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
80
Batch THF/MeTHF
Batch THF/MeTHF
Batch THF/MeTHF
Batch THF/MeTHF
Batch THF/MeTHF
Batch PhMe
1
40
1
40
1
40
54%
99.9:0.1
1
40
Demo
2
0.29
112.8
37 °C
82:18
60%
70%
99.9:0.1
99.4:0.6
Flow
THF/MeTHF
TMSCl
-75 °C 86:14
a) Determined by HPLC analysis.
Analysis of the 80 kg batch record revealed that the MeLi addition time had been almost 4 h instead
of the usual 1 h. Subsequent studies confirmed that the addition product was not stable in the
presence of MeLi under the reaction conditions. This caused us to consider continuous flow
conditions for the reaction, since under flow conditions the amount of time the product spent in
the presence of MeLi was significantly reduced and consistent. Continuous flow also offered
superior heat and mass transfer properties upon scale-up, which afforded consistent reaction
performance between the lab model and production.
A number of factors were studied in the development of a flow process capable of delivering
22•hemisulfate; the concentration of the feed solution of 21 was studied and it was found that
using 10 V of THF gave the best assay purity product at -70–80 °C and it was found that 2
equivalent of MeLi in 2-MeTHF was sufficient. Performing the reaction at -10–0 °C worsened the
dr (5:1), that performing the reaction at -78 °C and -40–50 °C both gave similar diastereoselectivity
(>6:1). The reaction was ultimately performed at -78 °C since it was easier to maintain a constant
ACS Paragon Plus Environment
17

Organic Process Research & Development
Page 18 of 38
1
2
3
4
5
6
temperature using a dry-ice bath. Indeed, application of the flow chemistry performed well at 113
kg scale, affording 70% yield and reasonable diastereoselectivity (86:14 dr) which was improved
to >99:1 through the salt formation. In total, this new process afforded 101 kg of 22•hemisulfate.
7
8
9
2.4 Issues encountered during Campaign 3
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
A further campaign to prepare 344 kg of 22•hemisulfate was required to support the production
of 100 kg of 3 for advancing clinical trials and so this sequence was executed again with some
changes to the process based on issues that were encountered in the previous campaigns and
incorporating the flow chemistry solution for the critical MeLi addition. The following paragraphs
outline those issues, the changes made in response and their implementation.
While most of the process was unchanged from campaign 2, a change was made to the
esterification (from AcCl to SOCl2) to afford 11•HCl to improve reaction rate (Scheme 9). As with
step 1, this afforded 11•HCl in high yield. While freebasing this HCl salt with K2CO3 can cause
hydrolysis of 11 back to 10 in MeOH, this issue was obviated in CH2Cl2 and so a solvent swap
from MeOH to CH2Cl2 was performed before the freebasing step. This ester can then be cleanly
reduced to 14 by portion wise addition of NaBH4 to control the evolution of hydrogen during the
reaction.
Br
O
Br
Br
O
1. K2CO3,
4 Batches 70-127 kg
77-82% yield
99% purity
CH2Cl2/H2O
SOCl2
MeOH
O
OH
OH
•HCl
2. NaBH4
NH
NH
14
•HCl
NH
10•HCl
99%ee
87-92%ee
11•HCl
Scheme 9: New process for preparation of 14.
While the new procedure for reduction avoided aluminium hydride reagents, some issues were
encountered in manufacture. The enantiomeric excess was eroded from ca. 98% to ca. 91% over
the course of esterification and ester reduction. It was not determined with certainty which step
was responsible, but the enantiomeric purity of 10 was easily ruled out as a possible cause and
laboratory studies were unable to replicate the erosion during the borohydride reduction step,
which may implicate the ester formation. Laboratory scale work showed that the ee of the material
could be improved in the final salt formation to afford 22•hemisulfate and so the ensuing steps
ACS Paragon Plus Environment
18

Page 19 of 38
Organic Process Research & Development
1
2
3
4
5
6
were performed per the normal process with no new modifications, until the final purification
(Scheme 10).
7
8
9
Br
Br
Br
1) CH2Cl2, K2CO3 basification,
solvent exchange to THF
H2SO4, EtOAc
OTBDPS
OTBDPS
OTBDPS
NH
NH
2) MeLi, 2Me-THF, -65 °C (flow)
Aqueous w-up,
Solvent exchange to EtOAc
1/2 H2SO4
N
HO
O
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
22•Hemisulfate
Campaign 3
HO
21•Oxalate
O
22
(75-80% de)
4 Batches ca. 110 Kg ea.
57-71% (2 steps)
344Kg produced
>99% purity
>99% de
>99:1 dr
Scheme 10: Production of 22•hemisulfate (Campaign 3).
Additional screening work in batch mode and flow determined that TMSCl was not necessary in
order to achieve high dr when MeTHF was used as solvent at low temperature. It is not understood
if this is purely a solvent effect or what the precise role of TMSCl was when used in conjunction
with MeLi. The third pilot plant campaign proceeded without TMSCl and used pure MeTHF as
reaction solvent. Four batches, each consuming ~110 kg of 21 were conducted; the crude reaction
profiles were generally consistent and the in-situ purity of 22 ranged from 84.5–87.1 HPLC area%
with the diastereomer present at 10.1–11.3 HPLC area% (dr = 7.0–8.6:1). Cryogenic conditions
were readily obtained at plant scale by submerging a plug flow reactor (PFR) in a cryogenic bath
and using dry ice and ethanol to cool the process (Figure 1).
Flow
Meter
Peristaltic
Pump C
320 ml/min
Balance
Manual Layer
Separations
Aqueous NH4Cl
2.94M
Flow
Meter
Flow
Meter
Pump D
Pump E
Peristaltic
Pump B
143.8 ml/min
Balance
MeLi in 2-MeTHF
0.60M
HPLC Sample
Collection
Cryogenic Bath
-70~-75oC
Diaphragm
Pump A
232 ml/min
Reactor
1.55L
V/Q = 4.1 min
316L SS
3/8" OD
Pre-Cool Coil
1.17L
V/Q = 3.1 min 27 helical elements
Static Mixer
0.319" ID
22 in 2-MeTHF
Balance
Continuous Quench
Aqueous Waste
21 in 2-MeTHF
0.195 M
0.28"ID
Figure 1. The process flow diagram of the experiment with continuous quench and phase
separation.
ACS Paragon Plus Environment
19

Organic Process Research & Development
Page 20 of 38
1
2
3
4
5
6
7
8
9
During the isolation of 22•hemisulfate it was found that low ee and low de were encountered in
the isolated solids. After a laboratory investigation, it was found that seeding the initial
crystallization from EtOAc with 2% (w/w) 22•hemisulfate was crucial to ensure a consistent
partial rejection of diastereomer and enantiomer. In the recrystallization from EtOAc/THF, partial
dissolution of the crude salt in THF followed by removal of the solid fraction by filtration was
implemented. Removal of the solid phase purged most of the enantiomer as scalemic material;
antisolvent addition and cooling finished the crystallization of high purity product. Laboratory
studies showed that removal of the enantiomer by crystallization appeared to have a kinetic
parameter and that aging of slurries of 22•hemisulfate resulted in reduced chiral purity as the
enantiomer began to crystallize out over time indicating that the process was at stochastic risk for
low ee if the enantiomer was present at significant levels. For the third manufacturing campaign,
22•hemisulfate was isolated in yields ranging between 57–71% (corrected for assay and purity)
with dr > 99.8:0.2 and ee >99.8%, which were similar quality parameters when compared to the
second manufacture. Yields suffered due to the required purge of enantiomer, but the yield from
the first batch in particular (57%) was reduced further due to challenges encountered during its
purge and potentially dryer hang-up. This new process allowed the production of 344 kg of
22•hemisulfate.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
3. Endgame synthesis.
During the discovery phase of this program, the final target compound was 3. However,
development of 3 was complicated by our inability to crystallize it under numerous experimental
conditions. Fortunately, we discovered a cocrystal form of 3 that not only possessed excellent
solubility but also met the criteria for clinical development. As a result, the 4-hydroxybenzoic acid
(HBA) cocrystal (3•HBA) has been carried forward into clinical trials and became the final target
compound for the process chemistry efforts.1
3.1 Discovery Chemistry Synthesis of 3.
The medicinal chemistry conversion of intermediate 17 to 3 was achieved in 5 steps and 30%
overall yield (Scheme 11). This sequence was potentially manageable at scale, but there were
opportunities to simultaneously shorten the route and remove some of the more challenging steps.
For further development, we therefore sought to install the side chain in such a way as to avoid the
ACS Paragon Plus Environment
20

Page 21 of 38
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
unnecessary use of ethyl acrylate (used in the conversion of 24 to 25) and MeLi (used in the
subsequent formation of 26), by using an alkene with the gem-dimethyl group already installed.
We also sought to replace the expensive rhodium catalyst that was employed to avoid
dehalogenation. Finally, the Heck coupling (24 to 25) was performed at a temperature considerably
in excess of the boiling point of the solvent.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
P(o-Tol)3 (24 mol%)
Pd(OAc)2 (5.4 mol%)
Et3N (3 equiv),
EtO
O
ethyl acrylate (3 equiv)
Br
OTBS
PyBOP (1.5 equiv)
Et3N (1.5 equiv)
DMF (15 vol)
OTBS
Br
OTBS
Cl
MeCN (1 M)
160 oC, 35 min
Cl
HO
Cl
+
NH
N
N
O
60%
83%
Cl
O
O
(1.1 equiv)
Cl
Cl
25
17
23
24
OH
OH
EtO
O
MeLi (3 equiv)
THF (0.175 M)
-78 o
Rh(cod)(diPrpf)BF4,
MeOH (30 vol)
H2 (100 psi), overnight
OH
TBAF (1.1 equiv.)
THF (0.1 M),
r.t., 40 min
OTBS
OTBS
C
Cl
Cl
Cl
N
O
N
N
80%
92%
81%
O
O
Cl
Cl
Cl
26
27
3
Scheme 11: Initial medicinal chemistry synthesis of 3.
3.2 Preclinical toxicology route to 3
In the medicinal chemistry route, the Heck reaction of aryl bromide 24 was performed after the
installation of the amide, which lead to several byproducts due to dechlorination or Heck coupling
in the amide portion of the molecule. We felt that if the order of steps could be changed to allow
introduction of the amide after this reduction, a less expensive catalyst might be employed. For
this reason, the Heck coupling was performed directly on 17 instead (Scheme 12). To shorten the
sequence and obviate the use of ethyl acrylate and MeLi, readily available and inexpensive 2-
methyl-3-buten-2-ol (28) was selected as the coupling partner.
Starting with conditions that were obtained for the same transformation on a different aryl halide
during another discovery program a small number of bases were surveyed (iPr2NEt, Cy2NMe and
K2CO3). When using freebase 17, Cy2NMe was preferred base, but if 17•HCl was used directly,
potassium carbonate was the superior base, affording 29 in reasonable yield. This Heck reaction
performed on compound 24 (Scheme 11) was considerably less efficient as significant
ACS Paragon Plus Environment
21

Organic Process Research & Development
Page 22 of 38
1
2
3
4
5
6
7
8
9
participation of the aryl chlorides was observed. As expected, the alkene reduction could be
achieved easily using palladium on carbon at modest pressure. The amide formation was facile
using 1,1’-carbonyldiimidazole (CDI) as a simple, inexpensive and non-allergenic14 coupling
agent with easily purged byproducts. Identical TBS deprotection conditions were then employed
to generate 3.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
OH
OH
Pd(OAc)2 (5 mol%)
SPhos (5 mol%)
K2CO3 (3.15 equiv)
DMF (3.5 mL/g)
5% Pd/C (10 mol%)
OTBS H2 (5 bar),
Br
OTBS
OTBS
Cl
EtOH,
rt, 4h
OH
HO
+
+
NH
NH
NH
O
64%
HCl
93%
Cl
29
23
(1.5 equiv)
28
30
17•HCl
OH
OH
TBAF (1.1 equiv.)
CDI (1.2 eq)
iPr2NEt (5 equiv.)
THF
OH
OTBS
THF (0.1 M),
PCy2
OMe
0 oC-rt, 40 min
Cl
MeO
Cl
N
N
72%
98%
O
O
Cl
Cl
SPhos
3
27
182 g Prepared
Scheme 12: Second generation discovery synthesis of 3.
3.3. Process Development Synthesis of 3.
The second-generation route was substantially more practical for operation at scale than the first-
generation medicinal chemistry route. Several points for development had been identified, and
these focused primarily on intermediate isolations and elimination of undesirable reagents and
solvents. From this point forward, 22•hemisulfate (see section 2.3.3) was used as the starting
material to fund all development work for the final steps.
While the Heck reaction was suitable for medium scale-work, optimization was merited for further
scale-up. Issues included low crude purity after the coupling, phase separation challenges caused
by the dark color of both liquid phases during workup, and a considerable amount of solid, which
further complicated the separation. By switching from DMF to toluene as the reaction solvent and
performing a freebasing step prior to the Heck reaction, a considerably cleaner crude material was
obtained. This enabled an effective crystallization of 32 (Scheme 13) after a solvent swap into n-
heptane. During the initial development phase, it had been observed that the reaction progressed
ACS Paragon Plus Environment
22

Page 23 of 38
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
faster with increased temperature. However, during the scale up the highest internal temperature
reached in the reactor was 109 °C. Any further attempts to increase the internal temperature by
increasing the jacket temperature without reactor pressurization resulted only in more vigorous
boiling. These conditions were applied in the first GMP production campaign to prepare 10 kg of
3•HBA. It became evident that the Heck reaction was slower than desirable on pilot plant scale
since the two batches required 63 and 77 h, respectively, of reaction time to achieve the end of
reaction specification for conversion.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In the next round of development, in order to speed the Heck coupling reaction rate, the process
was performed in a sealed pressure vessel to allow access to higher temperatures. After sealing,
the reactor was allowed to self-pressurize with all the pressure being supplied by the toluene vapor
pressure since the internal temperature exceeded the atmospheric pressure boiling point. It proved
possible to decrease the amount of the palladium acetate and SPhos from 0.03 equivalents to 0.015
equivalents, which served to increase the efficiency of the subsequent palladium scavenging
operation. Finally, the amount of the 2-methyl-3-buten-2-ol was decreased from 5 equivalents to
2.5 equivalents and the reaction volume was decreased from 9 to 4.5 V.
The hydrogenation conditions defined during the discovery program were largely suitable for
manufacturing scale work. Reaction optimizations decreasing the ethanol solvent amount from 12
to 8 L/kg olefin, use of lower catalyst loading (3 mol% vs. 10 mol%), and isolation of 33 after
solvent exchange from to ethanol to heptane to crystallize the product with subsequent isolation
by filtration.
The amide formation developed to this point was slow and often required additional charges of
CDI to reach completion. A survey of solvents suggested that THF should be retained, as THF
balanced speed of reaction, starting material solubility and TBDPS group retention. Reasonable
reaction times could be obtained by increasing the amide bond formation temperature from 45 ºC
to 65–70 °C. Additionally, the use of EtOAc during the work up with aq. NaOH was of concern
due to the potential hydrolysis of EtOAc. This was important because it was shown that a single
basic wash was not sufficient to fully hydrolyze the residual acylimidazole intermediate (31,
Figure 2) and remove residual 2,6-dichlorophenylacetic acid (23). Isolation of 34 by solvent
evaporation alone lead to significant foaming and potential contamination with 31. Substitution of
ACS Paragon Plus Environment
23

Organic Process Research & Development
Page 24 of 38
1
2
3
4
EtOAc with iPrOAc as the extraction solvent enabled the use of multiple basic washes of sufficient
5
6
duration to hydrolyze 31 and remove 23 while minimizing solvent hydrolysis.
7
8
Cl
N
N
9
O
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Cl
31
Figure 2: Structure of active amide intermediate during amide formation.
Since TBDPS-protected amide 34 was challenging to isolate, attention was paid to developing a
process where the iPrOAc workup solution could be telescoped directly into the deprotection step.
It was found that by adding TBAF•3H2O to the crude iPrOAc solution and heating to 60 °C, the
deprotection was complete in <1 h at laboratory scale. After aqueous workup, the crude iPrOAc
solution was treated with Biotage MP-TMT to obtain material with Pd content within specification.
During work to break the co-crystal 3•HBA with base, crystallization of the free form of 3 was
observed for the first time. Subsequently during laboratory-scale Pd scavenging trials, a product
solution obtained after removal of the scavenger by filtration through Celite was concentrated to
approximately 4 V and allowed to stand overnight at room temperature. During this time, 3
crystallized from solution, which provided an attractive potential impurity control element. The
crystalline free form has solubility of 18 mg/mL in iPrOAc. After further study, this crystallization
was included in the next GMP campaign. Initial discovery and development of the HBA co-crystal1
had been conducted with pure API in isopropanol/heptane as solvent but it was now possible to
precipitate the free form of 3 directly from the post-work up iPrOAc solution following the TBDPS
deprotection. This crystallization efficiently removed the TBDPSOH generated from the
deprotection but did not eliminate the tetrabutylammonium salts, so an aqueous wash was
employed to remove them before the co-crystal formation to deliver 3•HBA with suitable purity.
Low levels of related substances were adequately purged during the intermediate crystallizations
as well as the final 3•HBA crystallization.
With this development work in hand, the process (Scheme 13) was applied to the manufacture of
53.9 kg of 3•HBA. A third manufacturing campaign was performed delivering 119.8 kg of 3•HBA
in a very similar manner to the previous campaigns, with optimizations made either to reduce
reactor residence times or to reduce solvent volumes in reactions and workups. Compared to the
second manufacturing campaign, the Heck coupling was conducted at higher concentration (3.5
ACS Paragon Plus Environment
24

Page 25 of 38
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
L/kg toluene vs. 4.5 L/kg) and higher temperature (130–135 vs. 120–125 ºC). To improve
substrate solubility and reaction kinetics, the hydrogenation was performed in ethanol at 40–45 ºC
compared to the previous campaign (20–25 ºC). The amide bond formation step was conducted at
higher concentration in THF (4.0 L/kg vs. 5.5 L/kg), the telescoped deprotection step utilized less
i-PrOAc solvent (3 L/kg vs. 4 L/kg) and a more cost effective Pd scavenger was utilized (SiliaMetS
Thiol replaced Biotage MP-TMT). Finally, 3•HBA was wet milled to reduce the particle size.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
OH
OH
1. 1M K2CO3 (aq), PhMe
2. 28, SPhos, Pd(OAc)2,
K2CO3, 135 oC
Br
OTBDPS
OTBDPS
OTBDPS
Pd/C, H2 (5 bar),
EtOH, 40 oC
23, THF, CDI
then 33
NH •1/2H2SO4
NH
75-85%
NH
88%
22•Hemisulfate
OTBDPS
33
32
OH
OH
OH
1. TBAF.3H2O,
60-70 oC
2. SiliaMetSThiol,
1. THF, 4-Hydroxy benzoic acid
2. n-Heptane, wet milling
3. Filter, wash,dry
OH
COOH
OH
OH
O
Cl
Cl
Cl
N
N
88-90%
N
66-77%
O
Cl
O
Cl
Cl
3•HBA
120 kg produced
3
34
Scheme 13: Results from third clinical trial supply manufacturing campaign of 3•HBA.
5. Conclusions
In summary, a new process has been developed that enabled the study of mevidalen (3)
hydroxybenzoate in clinical trials for the potential treatment of Lewy body dementia. The overall
yield, step count and production quantities for these campaigns are outlined in Table 5. The initial
medicinal chemistry route to 3 was relatively long with some superfluous protecting group
manipulations leading to a low overall yield (Entry 1). These were eliminated in the route used to
deliver materials for preclinical toxicology experiments, resulting in a shorter and higher yielding
procedure (Entry 2). As 3 entered clinical study, the route was adjusted to use a more robust silicon
protecting group and to isolate two key intermediates as salts (21 and 22), allowing facile isolation
and impurity rejection. The manufacturing of this material to supply clinical trials was performed
over three separate campaigns: Campaign 1 (Entry 3) suffered from lower than expected yield due
to the large-scale cryogenic addition of MeLi to 21 not performing as expected on scale-up
ACS Paragon Plus Environment
25

Organic Process Research & Development
Page 26 of 38
1
2
3
4
5
6
7
8
9
(Section 2.3.3), which was remedied in Campaigns 2 and 3 by use of flow-chemistry. Campaign 3
(Entry 5) suffered some attenuation of overall yield due to the need for an additional processing
step to improve the chiral purity of 22•Hemisulfate after it was unexpectedly eroded during the
new reduction procedure (section 2.4).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 5: Campaign evolution summary
Entry
Campaign
Yield to
Yield to
Overall Total Steps
Amount
intermediated
3•HBAd
Yield
3.3%a
11.3%a
to 3•HBA
produced
1
2
Med Chem
11.1% (11 steps)b
27.3% (8 steps)b
29.7%a
42.0%a
16a
milligrams
182 g
Preclinical
Tox
12a
3
4
5
Campaign 1
Campaign 2
Campaign 3
22.2% (8 steps)c
33.6% (9 Steps)c
32.9% (9 Steps)c
45.0%
44.7%
46.3%
10.0%
15.0%
15.2%
14
14
15
18.8 kg
53.9 kg
119.8 kg
a Only free form (3) was synthesized not 3•HBA. For comparison, these routes are therefore inherently 1 step shorter without
concomitant reduction in yield. b) Intermediate = 17. c) Intermediate = 22•Hemisulfate d) Yield reported is an average of all
the batch yields for each step in that campaign.
Despite the significant progress that has been made in the development of a route capable of the
production of >100 kg of 3, this route suffers from a number of issues: (1) the synthesis route is
long and linear, resulting in a lengthy lead time for the API and reduced sourcing flexibility; (2)
sourcing large quantities of 1 is problematic as it is not readily available and must be produced on-
demand, resulting in long lead times; (3) installation of the C-1 methyl group occurs with good
diastereoselectivity but it cannot be performed directly and relies on the oxidation-addition
sequence; (4) the C-1 methyl group installation requires cryogenic conditions that were enabled
by continuous flow, but this reduces process portability; (5) multiple non-productive protecting
group manipulations are required; (6) unknown origin of the ee erosion between esterification and
borohydride reduction indicates further process understanding is needed; (7) the atom economy
and throughput for intermediate 22•hemisulfate is hindered due to the heavy bromine, silicon
protecting group, and salt form. Subsequent development efforts were undertaken to address these
ACS Paragon Plus Environment
26

Page 27 of 38
Organic Process Research & Development
1
2
3
4
5
problems and resulted in identification of improved alternative approaches to the THIQ core of 3.
15,16
6
7
8
Experimental Section
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Experimental procedures for Campaign 3.
1H NMR spectra are referenced to DMSO at 2.50 ppm and 13C NMR spectra to DMSO at 39.52
ppm.
(R)-5-bromo-1,2,3,4-tetrahydroisoquinoline-3-carboxylic acid Hydrochloride (10•HCl). 1
To a dry 3000 L glass-lined reactor was charged 1 (105.0 kg, 430.2 mol) and methanol (422 kg),
and the mixture heated to internal temperature of 62–65 °C. Thionyl chloride (63.4 kg, 532.9 mol)
was added and the mixture was refluxed at 62–65 °C for 2 h. After this time the temperature was
adjusted to 30–40 °C and a sample taken for analysis, which showed that <1% of 1 remained. The
mixture was concentrated to 1.01.5 volumes while maintaining the internal temperature below
40 °C. The mixture was diluted with CH2Cl2 (680 kg) and concentrated to 1.5–2.0 V while
maintaining the internal temperature below 40 °C. The mixture was diluted with CH2Cl2 (621 kg)
and cooled to below 10 °C. A solution of K2CO3 (120 kg, 868.3 mol) in water (360 kg) was added,
maintaining the internal temperature below 10 °C and held for 10 min. Methyl chloroformate (45.2
kg, 1.1 eq, 478.3 mol) was added over 20 min, maintaining the internal temperature below 25 °C,
and the mixture stirred at 1525 °C for 3 h. The phases were separated, and the organic phase was
filtered through Celite. To the filtrate was added CH2Cl2 (156 kg) and conc. HCl (56.2 kg) in order
to adjust the pH to 2–4 over 90 min. The phases were separated, and the organic phase was filtered
through Celite and concentrated to 1.5–2.0 V, maintaining the internal temperature below 35 °C.
Karl Fisher titration of the organic layer showed the water content was 0.5% (w/w). Acetic acid
(439.4 kg) was added over 20 min followed by paraformaldehyde (16.05 kg, 535.0 mol) over 20
min and the mixture was cooled to 010 °C. Then H2SO4 (387.6 kg) was added slowly, maintaining
the internal temperature below 25 °C, then held at 2025 °C for 20 h. A sample was taken for
analysis, which showed that less than 0.2% of 8 remained. The reaction mixture was cooled to
010 °C and water (1000 kg) was added, maintaining the internal temperature below 30 °C (5.5
h). EtOAc was added (463 kg) and the mixture was stirred between 69 °C for 30 min and the
ACS Paragon Plus Environment
27

Organic Process Research & Development
Page 28 of 38
1
2
3
4
5
6
7
8
9
layers were separated. The aqueous layer was extracted with EtOAc (350 kg) and the combined
organic layers were concentrated to afford a solution of 9 and 18 in EtOAc (406.2 kg). In a separate
reactor, conc. HCl (1301.2 kg) was heated to 50–60 °C and the EtOAc solution of 9 and 18 was
added over 4 h. The mixture was then heated to 96104 °C for 30 h and then cooled to 5060 °C.
h. A sample was taken for analysis, which showed that less than 0.2% 9 and 18 remained. The
mixture was then cooled to 20–25 °C and held for 5 h and then the solids collected by filtration.
The cake was washed with water (135.65 kg) and then slurried in CH2Cl2 (680 kg) at 2030 °C
for 5 h. The solids were collected by filtration and the wet cake washed with CH2Cl2 (138.5 kg).
The wet cake was dried under vacuum at 40-45 °C for 50 h to afford 10•HCl as a white solid
(106.4 kg of salt, 87.0% freebase assay, 361.5 mol freebase, 84.0% yield, 99.4% purity by HPLC,
98% ee). 1H NMR (400 MHz, DMSO-d6) 9.95 (br.s, 2H), 7.62 (d, J = 7.8 Hz, 1H) 7.33 (d, J = 7.3
Hz, 1H), 7.25 (dd, J = 7.8, 7.8 Hz, 1H), 4.47 (dd, J = 5.4, 11.2 Hz, 1H) 4.414.29 (m, 2H), 3.29
(dd, J = 5.4, 17.6 Hz, 1H), 2.97 (dd, J = 11.2, 17.1 Hz, 1H); MS (ES+) m/z 256.0, 258.0 (M+H);
DSC onset =291.4 °C, max = 297.6 °C.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
[(3R)-5-Bromo-1,2,3,4-tetrahydroisoquinolin-3-yl]methanol (14).1
A dry 3000 L glass-lined reactor was charged with 10•HCl (127 kg, 89.9% assay, freebase basis,
446.0 mol) and MeOH (1028 kg), then heated to an internal temperature of 62 °C. To the mixture
was added SOCl2 (82.2 kg, 690.2 mol) over 5 h, maintaining the internal temperature below 66
°C. After the addition, the mixture was held at 6365 °C for 10 h and then cooled to 29 °C. The
mixture was then concentrated to 12 volumes, maintaining the temperature between 1837 °C.
The mixture was diluted with CH2Cl2 (874 kg) and concentrated to 12 V, maintaining the
temperature between 59 °C. The mixture was then diluted with CH2Cl2 (878 kg) and a solution
of K2CO3 (123 kg, 890.0 mol) in water (650 kg) was added, maintaining the temperature between
1625 °C. The mixture was stirred for 20 min, maintaining the temperature between 1617 °C and
then filtered through Celite (20 kg). The layers were separated and the aqueous extracted with
CH2Cl2 (886 kg). The combined organic phases were washed with water (632 kg) and concentrated
to 1–2 volumes maintaining the internal temperature below 30 °C. The vessel was then charged
with MeOH (568 kg), concentrated to 12 V maintaining the temperature between 1030 °C, then
charged with MeOH (992 kg) and heated to 42 °C. NaBH4 (50.0 kg, 1321.7 mol) was added in
portions while maintaining the internal temperature between 4252 °C. The mixture was then
ACS Paragon Plus Environment
28

Page 29 of 38
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
cooled to 2224 °C for 2 h and an aliquot taken for analysis. This showed 4% 11 remained and so
the internal temperature was increased to 41 °C and additional NaBH4 (2.7 kg, 71.4 mol) was
added in portions, maintaining the internal temperature between 4346 °C. The mixture was then
cooled to 2224 °C for 2 h and an aliquot was taken for analysis. The mixture was concentrated to
5–6 vol. maintaining the internal temperature between 1626 °C and water (1300 kg) was added
over 4 h maintaining the internal temperature below 40 °C. Upon completion of addition, the
mixture was stirred between 2325 °C for 3 h and then cooled to 3–4 °C for 5 h. The solids were
collected by filtration and the cake washed with water (220 kg). The cake was then slurried with
CH2Cl2 (422 kg) and methylcyclohexane (756 kg) at 2030 °C for 5 h then filtered, washing with
methylcyclohexane (105 kg). Finally, the solids were dried under vacuum at 40–45 °C for 24 h to
afford 14 as a pale yellow solid (85.35 kg, 78%, 99% purity by HPLC, 91.6% ee, assay 99.8%).
1H NMR (400 MHz, DMSO-d6) 7.43-7.39 (m, 1H), 7.08-7.06 (m, 2H), 4.74 (ddd, J = 5.4, 5.4, 5.4
Hz, 1H), 3.95-3.85 (m, 2H), 3.53-3.48 (m, 1H), 3.45-3.40 (m, 1H), 2.85-2.79 (sext, J = 1.3 Hz,
1H), 2.67 (dd, J = 3.9, 17 Hz, 1H), 2.43 (s, 1H), 2.23 (dd, J = 10.5, 17 Hz, 1H). MS (ES+) m/z
242.0, 244.2 (M+H). 13C NMR (100 MHz, DMSO-d6) δ 139.6, 134.2, 129.5, 127.1, 125.5, 125.0,
65.0, 55.2, 48.1, 32.5; DSC onset = 155.4 °C, max = 159.1 °C.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(R)-5-bromo-3-(((tert-butyldiphenylsilyl)oxy)methyl)-3,4-dihydroisoquinoline
oxalate
(21•oxalate).
A dry 3000 L glass-lined reactor was charged with 14 (110 kg, 450.0 mol) followed by CH2Cl2
(748 kg) and imidazole (47.0 kg, 690.4 mol) and the contents were stirred between 1521 °C for
20 min. Then TBDPSCl (170.0 kg, 618.5 mol) was added in portions maintaining an internal
temperature below 30 °C. The mixture was then stirred between 2123 °C for 8 h and then cooled
to 3 °C and charged with Et3N (147.0 kg, 1452.7 mol). The mixture was held between 05 °C for
20 min and then charged with N-chlorosuccinimide (92.0 kg, 689.0 mol) in 3 portions maintaining
the internal temperature below 5 °C. After the addition, the mixture was stirred for 12 h, then an
aliquot taken for analysis which showed complete consumption of 20. The reactor was charged
with a solution of NH4Cl (1140 kg) in water (2084 kg) over 5 h 40 min and then stirred between
35 °C for 2.5 h. The layers were separated, and the organic phase was transferred to a different
3000 L glass-lined reactor and twice washed with water (562 kg & 552 kg). The organic layer was
ACS Paragon Plus Environment
29