Discovery of N-(4-{[5-Fluoro-7-(2-methoxyethoxy)quinazolin-4-yl]amino}phenyl)-2-[4-(propan-2-yl)-1 H-1,2,3-triazol-1-yl]acetamide (AZD3229), a Potent Pan-KIT Mutant Inhibitor for the Treatment of Gastrointestinal Stromal Tumors

Article abstract of DOI:10.1021/acs.jmedchem.8b00938

While the treatment of gastrointestinal stromal tumors (GISTs) has been revolutionized by the application of targeted tyrosine kinase inhibitors capable of inhibiting KIT-driven proliferation, diverse mutations to this kinase drive resistance to establish

Full text of DOI:10.1021/acs.jmedchem.8b00938

Subscriber access provided by University of South Dakota
Article
Discovery of N-{4-[(6,7-dimethoxyquinazolin-4-yl)oxy]phenyl}-2-[4-
(propan-2-yl)-1H-1,2,3-triazol-1-yl]acetamide (AZD3229), a potent pan-
KIT mutant inhibitor for the treatment of gastrointestinal stromal tumors
Jason Grant Kettle, Rana Anjum, evan barry, Deepa Bhavsar, Crystal Brown, Scott Boyd, Andrew
Campbell, Kristin Goldberg, Michael Grondine, Sylvie Guichard, Christopher Hardy, Tom Hunt, Rhys Jones,
Xiuwei Li, Olga Moleva, Derek Ogg, Ross Overman, Martin J Packer, Stuart Pearson, Marianne Schimpl,
Wenlin Shao, Aaron Smith, James Smith, Darren Stead, Stephen Stokes, Michael Tucker, and Yang Ye
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.8b00938 • Publication Date (Web): 11 Sep 2018
Downloaded from http://pubs.acs.org on September 11, 2018
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 39
Journal of Medicinal Chemistry
1
2
3
4
Discovery
of
N-{4-[(6,7-dimethoxyquinazolin-4-yl)oxy]phenyl}-2-[4-
5
6
(propan-2-yl)-1H-1,2,3-triazol-1-yl]acetamide (AZD3229), a potent pan-
7
8
9
KIT mutant inhibitor for the treatment of gastrointestinal stromal tumors
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
Jason G. Kettle,*† Rana Anjum,§ Evan Barry,§1 Deepa Bhavsar,§ Crystal Brown,§ Scott
Boyd,† Andrew Campbell,† Kristin Goldberg,† Michael Grondine,§ Sylvie Guichard,§2
Christopher J. Hardy,¶ Tom Hunt,† Rhys D.O. Jones,† Xiuwei Li,¥ Olga Moleva,† Derek
Ogg,¶3 Ross C. Overman,¶ Martin J. Packer,† Stuart Pearson,† Marianne Schimpl,¶ Wenlin
Shao,§ Aaron Smith,† James M. Smith,† Darren Stead,† Steve Stokes,† Michael Tucker,† Yang
Ye¥
†Oncology, IMED Biotech Unit, AstraZeneca, Unit 310 ꢀ Darwin Building, Cambridge
Science Park, Milton Road, Cambridge, CB4 0WG, United Kingdom.
¶Discovery Sciences, IMED Biotech Unit, AstraZeneca, Unit 310 ꢀ Darwin Building,
Cambridge Science Park, Milton Road, Cambridge, CB4 0WG, United Kingdom.
§Oncology, IMED Biotech Unit, AstraZeneca, 35 Gatehouse Park, Waltham MA 02451,
USA.
¥Pharmaron Beijing Co., Ltd. 6 Taihe Road BDA, Beijing 100176 P. R. China
1 Current address: H3 Biomedicine Inc. 3520, 300 Technology Square floor 5, Cambridge, MA 02139, USA.
2 Current address: Forma Therapeutics Inc. Forma Therapeutics, 500 Arsenal St, Watertown, MA 02472, USA.
3 Current address: Peak Proteins Ltd., 3G48 Mereside, Alderley Park, Macclesfield, Cheshire, SK10 4TG, UK.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 39
1
2
3
4
5
6
7
8
9
Abstract While the treatment of Gastrointestinal Stromal Tumors (GIST) has been
revolutionised by the application of targeted tyrosine kinase inhibitors capable of inhibiting
KITꢀdriven proliferation, diverse mutations to this kinase drive resistance to established
therapies. Here we describe the identification of potent panꢀKIT mutant kinase inhibitors that
can be dosed without being limited by the tolerability issues seen with multiꢀtargeted agents.
This effort focussed on identification and optimisation of an existing kinase scaffold through
the use of structureꢀbased design. Starting from a series of previously reported
phenoxyquinazoline and quinoline based inhibitors of the tyrosine kinase PDGFRα, potency
against a diverse panel of mutant KIT driven Ba/F3 cell lines was optimised, with a particular
focus on reducing activity against a KDR driven cell model in order to limit the potential for
hypertension commonly seen in second and third line GIST therapies. AZD3229
demonstrates potent single digit nM growth inhibition across a broad cell panel, with good
margin to KDRꢀdriven effects. Selectivity over KDR can be rationalised predominantly by
the interaction of water molecules with the protein and ligand in the active site and its kinome
selectivity is similar to the best of the approved GIST agents. This compound demonstrates
excellent crossꢀspecies pharmacokinetics, shows strong pharmacodynamic inhibition of
target, and is active in several in vivo models of GIST.
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
Introduction Gastrointestinal Stromal Tumors (GIST) are soft tissue sarcomas of the
gastrointestinal (GI) tract that are driven predominantly through aberrant signalling of the
protoꢀoncogene cꢀKIT.1 Following surgical resection, fiveꢀyear survival rates vary from 35 ꢀ
65% depending on the size of tumor, mitotic potential, and where it is located.2 Whilst
incidence is relatively rare, estimated to be of the order of 1.5 patients per 100,000 population
annually3 (annual incidence in the United States is estimated at 3800),4 GIST is still the most
common mesenchymal tumor of the GI tract, with the majority harbouring activating
ACS Paragon Plus Environment

Page 3 of 39
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
mutations of this receptor tyrosine kinase.5 Such mutations are diverse and heterogeneous,
spanning the kinase domain of KIT with single point amino acid transpositions in the ATP
binding pocket, sequence deletions in exon 11, insertions in exon 9, with concurrent primary
and secondary mutational sites being common. A smaller subset of GIST tumors are instead
driven by mutations in the related tyrosine kinase, plateletꢀderived growth factor receptorꢀα
(PDGFRα). The D842V mutant of PDGFRα in particular, is equivalent to the KIT D816X
mutations, which are refractory to current therapies.6
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
Beyond surgical removal, treatment of GIST has been revolutionised by the application of
targeted tyrosine kinase inhibitors capable of inhibiting KITꢀdriven proliferation. Imatinib 1,
an inhibitor of both cꢀKIT and PDGFRα, was initially approved in 2002 and granted
accelerated approval in 2008 as first line adjuvant treatment following surgery.7 Clinical trials
have demonstrated a 5ꢀyear overall survival rate of 92% following 3ꢀyear therapy, compared
with 82% for those taking the drug for one year. However, up to 20% of GIST patients
exhibit primary resistance to imatinib, and many patients taking imatinib will eventually
relapse due to secondary mutations which are unresponsive to therapy, although in some case
higher doses of the drug may prove effective.8 In 2006, the FDA approved multiꢀtargeted
kinase inhibitor sunitinib 2 as a second line treatment for GIST patients who are refractory to,
or intolerant of imatinib therapy. In these patients, timeꢀtoꢀtumor progression of sunitinibꢀ
treated patients was superior to that of placeboꢀtreated patients.9 A second multiꢀtargeted
kinase inhibitor, regorafenib 3, was approved as third line therapy in 2013 for patients with
GIST that is not amenable to surgery and is refractory to both imatinib and sunitinib.
Approval was again based upon improved progression free survival (PFS) versus placebo.10
Despite the availability of these targeted agents, many clinically observed mutants of cꢀKIT
remain unresponsive, and there remains a significant unmet need in this disease for agents
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 39
1
2
3
4
5
6
7
8
9
with a much broader mutant KIT inhibition profile. The tolerability of sunitinib and
regorafenib therapy, whilst clinically manageable, may also present challenges.11 Toxicities
associated with their multiꢀtargeted kinase inhibition profile can lead to dose reductions and
drug holidays, which may in theory exacerbate emergence of further resistant clones. Potent
inhibition of VEGFR in these agents is associated with significant highꢀgrade hypertension,
amongst other significant toxicities.12
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
1
2
3
4
5
Figure 1. Approved GIST therapies and agents in clinical trials. Imatinib 1, sunitinib 2, regorafenib 3 are
approved as first, second and third line treatments respectively. Ripretinib (DCCꢀ2618) 4 and avapritinib (BLUꢀ
285) 5 are currently undergoing clinical trials in GIST patients.
Chemistry The synthesis of inhibitors 15 and 16 (Table 2) and related analogues 17 – 21
(Table 3) have been previously described.13, 14, 15, 16 A representative synthesis of compound
44 is shown in Scheme 1, with detailed experimental conditions for additional compounds
available as Supplementary Information. Accordingly, copperꢀcatalysed cycloaddition of
ethyl azido acetate 6 with 3ꢀmethylbutꢀ1ꢀyne gives triazole 7 which was hydrolysed to
ACS Paragon Plus Environment

Page 5 of 39
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
carboxylic acid 8. Amide coupling of 8 with a BOCꢀprotected bisꢀaniline gave 9 and removal
of the protecting group afforded aniline 10. Starting from difluorocyanophenol 11, alkylation
gave methoxyethyl ether 12, and displacement of one fluoro atom with ammonia under
microwave heating gave aniline 13. This was converted to the dimethylformimidamide
cyclisation precursor 14 and condensation of this with prepared aniline 10 in acetic acid gave
44, synthesising the quinazoline and introducing the C4 group concomitantly.
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 1. Synthesis of compound 44.
(i)
(iii)
R = Et
R = H
R = BOC
6
7
8
9
(ii)
(iv)
R = H
10
(vi)
(viii)
R = H
R = CH CH OMe
R = NH
2
11
12
13
14
44
(v)
(vii)
R = NCH=NMe
2
2
2
Reagents and conditions: (i) CuI, 3ꢀmethylbutꢀ1ꢀyne, Et3N, CH3CN, 20oC, 3 days, 90%; (ii) LiOH, water, THF,
20oC, 1.5 h, 100%; (iii) tertꢀButyl Nꢀ(4ꢀaminophenyl)carbamate, HATU, DIPEA, DMF, 20oC, 16 h, 57%; (iv)
4M HCl in dioxane, MeOH, DCM, 20oC, 3 h, 89%; (v) 1ꢀBromoꢀ2ꢀmethoxyethane, K2CO3, DMF, 85oC, 5 h,
93%; (vi) aq. NH3, iꢀPrOH, 100oC, 13 h, 87%; (vii) 1,1ꢀDimethoxyꢀN,Nꢀdimethylmethanamine, 80oC, 2 h,
100%; (viii) 10, acetic acid, 60oC, 35 mins, 58%.
Results and discussion In order to screen for broad spectrum mutant cꢀKIT inhibitors with
low potential for hypertension, we established a panel of Ba/F3 cell lines expressing a variety
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 39
1
2
3
4
5
6
7
8
9
of clinically relevant secondary mutants of cꢀKIT, spanning the ATP binding sequence
(V654A, T670I gatekeeper) and the Aꢀloop (D816H, D820A, D823Y and others) all on a
background of a primary exon 11 deletion (of amino acids 557 and 558), together with a line
expressing TelꢀKDR to monitor selectivity over KDR (VEGFR2). In this latter line, KDR is
expressed as a fusion protein with a Tel sequence at the Nꢀterminus, resulting in constitutive
kinase activity. We chose exon 11 deletion as the primary mutation and the backbone for all
the secondary mutations since twoꢀthirds of GIST tumors harbor mutations in exon 11. Exon
11 deletions are associated with a shorter progressionꢀfree and overall survival in comparison
to the other exon 11 mutations. In particular, deletions involving codon 557 and/or codon 558
are associated with malignant behaviour.17,18 Proliferation of these cell lines is dependent
upon the introduced construct for growth, and so to monitor for nonꢀspecific effects on
viability in these cells, a parental Ba/F3 line (no introduced construct) is run in parallel.
Engineered cell lines including Ba/F3 have been shown previously to faithfully mimic
clinical efficacy of various agents in GIST.19,20 It is notable that all approved therapies are
type II kinase inhibitors, binding to the DFGꢀout kinase form, and in particular, cꢀKIT is one
of only a small number of kinases that prefers to exist in the DFG out form in its ground
state.21 Additionally cꢀKIT has an unusual activation profile in that the activation loop seems
not to require presence of phosphate22 and as type II inhibition in general is difficult to
reproduce in biochemical assays, given reliance on the activation state, we viewed a cellular
cascade as preferred for hit finding activities. Profiling of the various approved therapies in
these lines is informative (Table 1). In line with clinical experience, 1 shows only modest
activity against exon 11 deletion + V654A and D816H, and no activity at all against the
gatekeeper T670I mutation. Sunitinib 2 is potent against V654A and T670I, but has weak
activity against D816H. Regorafenib 3 is potent against the T670I mutant but is rather
modest against the others, and both 2 and 3 show potent inhibition of the TelꢀKDR line
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
ACS Paragon Plus Environment

Page 7 of 39
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
consistent with their potent effect on VEGFR2. Beyond these approved agents, additional
compounds are currently undergoing clinical trials in GIST patients, aimed at targeting
mutants of KIT not currently addressed in the clinic, with secondary D816H being a
particularly refractory example. Ripretinib (DCCꢀ2618), 4 is a potent panꢀKIT and PDGFR
inhibitor from Deciphera Pharmaceuticals and shows broad inhibition across this mutant
panel.23 It is also a potent inhibitor of the TelꢀKDR line however and indeed grade 3/4
hypertension has been reported in a recent phase 1 clinical study of this compound.24
Avapritinib (BLUꢀ285), 5 from Blueprint Medicines is a potent inhibitor of KIT exon 17
mutations including D816 variants and also PDGFR.25 In this panel it potently inhibits exon
11 deletion + D816H although is weaker against V654A and T670I. It has good selectivity
for TelꢀKDR, but a moderate signal in the parental line may reflect other kinase activities for
this agent.
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 1. Growth inhibition potency of GIST treatments in KIT mutant Ba/F3 lines and effect on TelꢀKDR.
Ba/F3 GI50 (ꢀM)
exon 11 del +
V654A
exon 11 del exon 11 del
ID
Phase
Parental
>10
KDR
>10
+D816H
+ T670I
1
Approved 1st line
Approved 2nd line
Approved 3rd line
0.393
0.006
0.231
0.031
0.292
0.535
>10
Imatinib
2
4.789
9.953
8.037
4.075
0.398
0.290
0.013
0.017
0.005
0.033
0.037
0.860
0.033
0.114
0.098
4.952
Sunitinib
3
Regorafenib
Ongoing clinical
trials
4
Ripretinib
Ongoing clinical
trials
5
Avapritinib
All activity data are reported in ꢀM and are the mean of at least n=2 determinations and have SEM within 0.2
log units.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 39
1
2
3
4
5
6
7
8
9
We have previously reported on efforts to develop type II quinazoline based inhibitors of
both PDGFR and VEGFR that culminated in the discovery of AZD2932, 15.26 Follow up
work has also been described aimed at generating a PDGFRꢀselective agent from this
template which led to quinoline 16 (compound 23 in this referenced work).27 We reasoned
that such inhibitors of PDGFR might show inhibition of some KIT mutants due to the close
homology of KIT and PDGFR, and the precedent established by compounds like 1. What was
not clear however is whether the broad and selective profile we desired would be achievable
– the imatinib precedent also establishes that potent PDGFR inhibitors do not necessarily lead
to broad activity against KIT mutants. Testing of 15 in this panel (Table 2) highlighted that
AZD2932 did indeed show the desired broad spectrum of inhibition, but that it also carried
potent KDR activity, as it was designed to (encouragingly, no adverse signal was seen in the
parental line, suggesting little off target activity). Conversely, the PDGFꢀselective compound
16, despite weak effects on the KDR cell line, was a poor inhibitor of KIT mutants and also
showed some activity in the parental line. The challenge then was to understand if the potent
activity of analogues like 15 could be combined with good selectivity against KDR seen for
16, and work that led to the latter had highlighted the importance in the nature of the central
phenoxy ring and terminal heterocycle as being important drivers of KDR selectivity.
Compound 17 in which the phenoxy group is replaced with an anilino linker loses potency
against all KIT mutants, and also KDR (Table 3). The drop in potency against secondary Aꢀ
loop mutant D816H is almost identical to the drop in KDR potency (37ꢀ versus 39ꢀfold) in
this instance. Adding a metaꢀmethoxy group to the phenoxy ring in 18 results in maintenance
of KIT potency with a modest improvement in KDR selectivity, in line with SAR previously
reported for this series (selectivity of D816H over KDR is 17ꢀfold, compared with almost
parity for 15). Conversion of the central phenyl ring of 18 into a metaꢀpyridine 19 results in
significant diminishment of KDR potency, and greater selectivity over D816H at 45ꢀfold,
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
ACS Paragon Plus Environment

Page 9 of 39
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
although this does come with weaker overall D816H potency at 176 nM. Transposition of
this into its quinoline analogue 20 results in restoration of KIT mutant potency and
concomitant enhancement of selectivity over KDR (145ꢀfold). The final compound in this
matrix, 21 combines the potent quinoline hinge binder with the metaꢀmethoxy phenol linker
of 18, to give single digit nanomolar inhibition of the three KIT lines, although in this
instance KDR selectivity has been substantially impacted. Testing of additional compounds
available from the original PDGFR program highlighted that other substituted end ring
heterocycles such as the 3ꢀmethylpyrazole found in 16 was not tolerated in cꢀKIT, and so
could not be used to drive further levels of KDR selectivity as had been done previously (data
not shown).
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. Growth inhibition potency of literature AZ PDGFR inhibitors in KIT mutant Ba/F3 lines and effect on
KDR.
15
16
Ba/F3 GI50 (ꢀM)
exon 11 del +
V654A
exon 11 del +
D816H
exon 11 del +
T670I
ID
Parental
>10
KDR
0.093
2.674
15
0.012
0.685
0.070
0.786
0.004
1.918
AZD2932
4.289
16
All activity data are reported in ꢀM and are the mean of at least n=2 determinations and have SEM within 0.2
log units.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 39
1
2
3
Table 3. SAR of KIT mutant Ba/F3 lines and effect on TelꢀKDR selectivity
4
5
6
7
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
Ba/F3 GI50 (ꢀM)
exon 11 del exon 11 del exon 11 del
ID
17
18
19
20
21
X
N
Y
CH
CH
N
Z
NH
O
R
Parental
>10
KDR
3.595
0.442
8.030
3.924
0.077
+ V654A
+ D816H
+ T670I
H
0.337
2.592
0.078
N
MeO
MeO
MeO
MeO
>10
0.008
0.021
0.013
0.006
0.026
0.176
0.027
0.008
0.012
0.060
0.061
0.007
N
O
>10
CH
CH
N
O
>10
CH
O
>10
All activity data are reported in ꢀM and are the mean of at least n=2 determinations and have SEM within 0.2
log units.
In order to rationalise the observed selectivity margins for compounds 17ꢀ21, we obtained Xꢀ
ray crystal structures for compound 18, bound to the kinase domain of both wild type KIT
and KDR (see supplementary data for details). We did not attempt to crystallise in multiple
KIT mutants, since we observed broad mutant selectivity over KDR and therefore reasoned
that selective motifs, if apparent, would be revealed by contacts in the Xꢀray construct of
KIT. There is high homology between KIT and KDR wild type sequences; the sequences are
56% identical in the kinase domains used for crystallography. A key difference between KIT
and KDR lies in the gatekeeper residue, which is T670 in KIT and V916 in KDR. Other
differences are more conservative, such as for hinge residues, which are Y672 (KIT) and
ACS Paragon Plus Environment

Page 11 of 39
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
F918 (KDR). The binding modes of 18 in the overlaid kinase domains are shown in figure 2.
The binding poses are remarkably similar in the two kinases, with very few contacts to nonꢀ
conserved residues. The orientation of this ligand is as expected for a Type 2 kinase inhibitor.
The isopropyl group binds into the DFG pocket, with no distinction between KIT and KDR,
and the quinazoline nitrogen binds to the hinge. The two methoxy groups project out into
solvent. We speculate that the modest selectivity appears to arise from a less favourable
orientation of the central phenyl ring with respect to the gatekeeper V916 side chain of KDR,
relative to T670 in KIT. The orientation of the phenyl is dictated by the methoxy substituent
and hinders a binding mode which could optimise packing to V916 of KDR. The bound
ligand creates a poorly solvated cavity between ligand and gatekeeper, which can be
ameliorated in KIT by the more polar Thr side chain, with a favourable interaction between
quinazoline nitrogen and side chain hydroxyl.
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
Figure 2. Kinase domains of KIT and KDR, overlaid on the binding site of Compound 18. The KIT structure is
coloured yellow with labels in bold typeface, the KDR structure in blue with normal typeface. Side chains are
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 39
1
2
3
4
5
shown for the DFG motif and the catalytically important GluꢀLys pair, as well as the gatekeeper (GK) residue
(PDB codes: 6GQJ and 6GQO).
6
7
8
9
We were motivated to explore the impact of reversing the acetamide linker in this series and
so made reversed acetamide matched pairs to simple analogues available from the original
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
PDGFR program (Table 4). Compound 22 is the N
ꢀethylpyrazole analogue of 15,16,26 and its
reversed acetamide counterpart 23 encouragingly had a largely similar profile. Selectivity
against KDR is moderately enhanced in the reversed acetamide, 6ꢀfold from roughly 1:1 over
D816H, while affinity for the T670I gatekeeper is most severely impacted (down 20ꢀfold),
activity against V654A and D816H are largely similar. Analogue 24, in which the central ring
is appended with a methoxy group, is a first indication however that the SAR that was
apparent in the original acetamide series does not transfer when the acetamide is reversed.
Whereas previously a 17ꢀfold improvement in selectivity was noted, here selectivity is
unchanged, as is potency across all the mutants – methoxy on the central ring appears to offer
no advantage. A similar breakdown in SAR was seen for introduction of a central pyridine
ring (Table 5). In the quinoline matched pairs 25 (phenyl) and 26 (pyridine), introduction of
nitrogen causes a significant drop in all KIT potencies except T670I, and although KDR
potency is weaker, selectivity is unchanged.
Table 4. Impact of reversing the acetamide linker on KIT and KDR potency.
Acetamide
Reversed acetamide
ACS Paragon Plus Environment

Page 13 of 39
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Ba/F3 GI50 (ꢀM)
exon 11 del exon 11 del exon 11 del
ID
22
23
24
Amide
R
H
Parental
>10
KDR
0.680
1.615
2.551
+ V654A
+ D816H
+ T670I
Acetamide
0.044
0.488
0.007
Reversed
acetamide
H
>10
0.021
0.078
0.250
0.251
0.141
0.086
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
Reversed
acetamide
MeO
>10
All activity data are reported in ꢀM and are the mean of at least n=2 determinations and have SEM within 0.2
log units.
We obtained a matched pair of Xꢀray structures for compound 23, as shown in figure 3. In
this case, we observed ligand binding modes which are slightly offset in the vicinity of the
hinge and gatekeeper, with the pendant ethyl group of the pyrazole ring coincident in the two
binding sites. This suggests that interaction in the DFG pocket is dominant for this scaffold,
but that selectivity arises once again from the gatekeeper pocket. Since 23 is a quinazoline,
the N3 atom is available to chelate a water molecule, which is now visible in the KIT
structure (shown as a red sphere). There is no water observed in KDR here, our assumption
being that this would be very unstable when packed against the V918 side chain. Instead, in
KDR, the ligand partially occupies the solvent pocket in preference to water, weakening both
the hinge interaction and the packing against the gatekeeper residue. We assumed that a
similar solvation stabilisation of 18 is possible, but this was not observed in our Xꢀray
structures (figure 2). The nitrogen position in the reversed acetamide precludes a direct
hydrogen bond with the glutamic acid chain (E640/E885), in contrast to that observed in the
normal acetamide. This region of the complex remains quite polar however, so that a
crystallographic water is observed to bind in both CKIT and KDR, as shown in Figure 3. The
water appears to mediate an interaction between acid side chain and ligand. The water
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 39
1
2
3
4
5
positions are almost coincident in CKIT and KDR, so that we could not use this interaction to
explain selectivity between the two proteins.
6
7
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
Figure 3. Kinase domains of KIT and KDR, overlaid on the binding site of reversed acetamide compound 23.
KIT protein ribbons are coloured by sequence identity, with blue residues identical and yellow nonꢀidentical to
KDR. The binding mode is offset between the two kinases, but there are still very few contacts to nonꢀconserved
residues. Water molecules are shown as red speheres in KIT and blue spheres in KDR). A water molecule is
observed in the KIT structure, but is absent from the KDR structure. Selectivity appears to arise from less
favourable orientation of the central aromatic ring against the V916 side chain of KDR, relative to the solvated
T670 side of KIT. In addition, the hinge interaction in KDR seems to be less optimal than for KIT. (PDB codes:
6GQK and 6GQP)
Table 5. Impact of reversing a pyridine linker on KIT and KDR potency.
ACS Paragon Plus Environment

Page 15 of 39
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
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
Ba/F3 GI50 (ꢀM)
exon 11 del + exon 11 del + exon 11 del +
ID
25
26
Y
CH
N
Parental
>10
KDR
0.082
0.761
V654A
D816H
T670I
0.026
0.047
0.007
>10
0.150
0.518
0.008
All activity data are reported in ꢀM and are the mean of at least n=2 determinations and have SEM within 0.2
log units.
With previous SAR unsuited to driving the greater levels of potency and selectivity required,
we turned attention to varying the endꢀring of the reversed acetamide using a series of
aromatic acetic acids available through an in house building block initiative.28 Table 6 shows
the impact of varying the nature of the 5ꢀmembered heteroaromatic portion of the molecule,
and here we focused on finding the optimal balance of exon 11 deletion + D816H potency
(since this appears the most challenging mutant to inhibit with this template) and KDR
selectivity, whilst keeping LogD as low as possible. Compound 27, serves as a useful
benchmark for these changes as it is the reversed acetamide version of analogue 15 – potency
and lipophilicity are good, but KDR selectivity is modest. Switching the isopropyl group to a
difluoromethyl 28 gives a largely similar profile. Reversing the pyrazole so that connection to
the acetamide is via an Nꢀlink as in analogues 29 and 30 provides compounds with improved
KDR margin but comes at the expense of increased lipophilicity. The last pyrazole in this set
31 with a free NH gives a very potent D816H inhibitor, reasonable selectivity but again with
higher than desirable LogD of 3.6. A Cꢀlinked imidazole 32 sees a drop in activity and
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 39
1
2
3
4
5
6
7
8
9
selectivity, whereas linking through nitrogen as in 33 gives a potent and selective inhibitor. A
thiadiazole 34 is potent but has only 10ꢀfold margin to KDR and LogD of 3.3. We also
examined a series of 1,2,3ꢀtriazoles 35 – 37. The pick of these appears to be triazole 35,
which combines good potency (19 nM) and selectivity with an acceptable LogD of 2.9.
Triazole 36 is less potent and triazole 37 less selective than 35. In related 1,2,4ꢀtriazoles,
compound 38 sees potency significantly diminished, and analogue 39, although an
improvement, is not superior to any 1,2,3ꢀtriazole. Completing this set, the tetrazole 40 shows
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 reasonable balance of potency, selectivity and lipophilicity. From this work
Nꢀlinked
triazole 35 emerged as the most favourable endꢀring, although further increases in selectivity
were desired.
Table 6. Exon 11 deletion + D816H KIT mutant potency and KDR selectivity for a series of aromatic
acetamides.
exon 11 del D816H
Fold KDR
selectivity
ID
27
R
LogD2
Ba/F3 GI50 (ꢀM)1
0.021
0.087
0.021
5.9
4.4
2.9
2.7
3.8
28
29
11.4
ACS Paragon Plus Environment

Page 17 of 39
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
0.035
0.009
23.5
15.6
3.5
3.6
30
31
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
0.240
0.021
0.026
0.019
0.111
0.075
1.369
0.175
0.042
4.6
34.1
9.9
2.8
3.3
3.3
2.9
2.6
2.9
2.3
2.6
3.0
32
33
34
35
36
37
38
39
40
32.2
12.6
9.3
3.9
15.9
15.7
1 All activity data are reported in ꢀM and are the mean of at least n=2 determinations and have SEM within 0.2
log units. 2 Measured LogD at pH7.4
.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 39
1
2
3
4
5
6
7
8
9
The notable gains in potency and selectivity for these heterocyclic end groups prompted us to
solve a further Xꢀray matched pair, this time for compound 35. As seen in figure 4, the
binding mode at the hinge and gatekeeper is less offset than for 23 (figure 3), but this appears
to be due to a different orientation in the DFG pocket. The pendant isopropyl group packs
against the V898 side chain of KDR but projects further from the pocket in KIT due to a
larger I653 side chain at the equivalent position. This results in a significant offset of the
triazolo group between KIT and KDR and a different orientation for the isopropyl group. The
triazole dipole is oriented towards bulk water, as we would expect for a broadly hydrophobic
pocket and consequently solvent mapping suggests that the triazolo motif occupies a more
favourable position in KIT relative to KDR.29 In KDR, it appears to be buried too far into the
pocket, to optimise packing against V898, precluding effective solvation of the exposed
dipole. This also explains why compound 27 is less selective, since the pyrazole end group
can present a hydrophobic atom in the KDR pocket. Combining isopropyl with a heterocycle
with significant dipole, such as triazolo, gives a binding orientation which is optimal for
solvation in the KIT pocket and subꢀoptimal for KDR. When coupled with a selective
gatekeeper motif and the bound water (also visible in figure 2), we then see a significant
selectivity gain.
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
ACS Paragon Plus Environment

Page 19 of 39
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
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
Figure 4. Kinase domains of KIT and KDR, overlaid on the binding site of compound 35. KIT protein is
coloured by sequence identity, with blue residues identical and yellow nonꢀidentical to KDR. The binding mode
at hinge and gatekeeper is less offset than in previous examples, but attention now shifts to nonꢀconserved
residues in the DFG pocket. The isopropyl group clashes with I653 in KIT, but is accommodated by the smaller
V898 side chain of KDR. The triazolo motif is more buried in KDR as a result and we hypothesise that this
results in a desolvation penalty which increases affinity to KIT for this ligand and for this general DFG pocket
motif. (PDB codes: 6GQL and 6GQQ).
We next turned our attention to examining substituents at the quinazoline 5ꢀposition, since
this is a region that projects towards the ribose pocket of ATP, an area of fruitful optimisation
in quinazolineꢀbased kinase inhibitors historically.30 Compound 35 is a potent inhibitor of all
three KIT mutants (Table 7). Migration of the 6ꢀmethoxy group to the 5ꢀposition 41 does not
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 39
1
2
3
4
5
6
7
8
9
change the profile a great deal – a modest 2 to 3ꢀfold drop in potencies are observed across all
cell lines, and certainly KDR selectivity is not further enhanced. Despite this, we explored the
impact of switching the 4ꢀphenoxy linker back to a 4ꢀanilino linker with compound 42. We
had seen earlier in the normal amide series, compound 17 with an anilino linker lost
significant activity against KIT mutants, although we speculated that the potential lone pair
clash between the adjacent oxygen atoms (at C4 and C5) in 41 might impact the optimal C4
substituent vector, and that reversion to an NH linker, which could internally hydrogen bond
to the 5ꢀmethoxy group, might be more favourable. In the event, the aniline linked 42 was as
effective at inhibiting KIT mutants as its phenoxy counterpart, but with much improved
selectivity over KDR (113ꢀfold compared with 24ꢀ and 32ꢀfold for 41 and 35 respectively). It
is striking that two structural changes that individually lead to a worsened profile, when
combined together are able to drive a significant benefit. Exploration of alternative
substituents at C5 led to 5ꢀfluoro analogue 43 which maintains KIT potency and leads to even
greater KDR selectivity of at least 400ꢀfold. Fluorine was selected as an alternative to
methoxy that could, potentially at least, form an internal hydrogen bond to the C4 NH. This
compound however unexpectedly suffered from adverse physicochemical properties, in
particular high plasma protein binding of 99.7% bound, an issue that was directly addressed
by variation of the C7 substituent. Compound 44, containing a C7 methoxyethoxy group was
determined to have the optimal balance between KIT mutant potency and KDR selectivity
from this work and was selected for further profiling. It demonstrates very potent growth
inhibition across KIT mutant cell lines with good KDR margin – not quite as selective as 43,
but significant improvements in physicochemical and ADME properties (Table 8). Despite a
low solubility consistent with its chemical structure and neutral chemotype, permeability in
Caco2 cells is very high. This, coupled with its low in vitro clearance in hepatocytes leads to
excellent observed in vivo pharmacokinetics in preclinical species. Bioavailability is high,
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
ACS Paragon Plus Environment

Page 21 of 39
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
and clearance low across all of mouse, rat and dog. Volume of distribution is low consistent
with the neutral structure. A structure of 44 bound in KIT was obtained and shows features
consistent with the preceding Xꢀray structures (figure 5). There is a bound water at the
gatekeeper, promoting selectivity, and the triazolo group binds in a similar place in the DFG
pocket as was seen for compound 35. The extended C7 side chain protrudes from the active
site out into 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
Table 7. SAR of 5ꢀsubstituted quinazolines.
35
Ba/F3 GI50 (ꢀM)
exon 11 +
del V654A del D816H
exon 11 +
exon 11 +
del T670I
ID
35
41
42
43
44
Z
R1
R2
Parental
>10
KDR
0.612
1.222
5.408
>10
0.003
0.007
0.003
0.002
0.003
0.019
0.050
0.048
0.025
0.009
0.017
0.055
0.107
0.068
0.016
O
MeO
MeO
F
Me
Me
>10
NH
NH
NH
>10
Me
>10
F
MeO(CH2)2
>10
1.378
All activity data are reported in ꢀM and are the mean of at least n=2 determinations and have SEM within 0.2
log units.
Table 8. Detailed ADME characteristics of compound 44.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 39
1
2
3
H
N
O
N
4
5
6
N
F
H N
N
7
8
9
N
O
N
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
Parameter
Value
3.0
LogD pH7.41
Solubility pH6.5 (ꢀM)2
6
Permeability Caco2 A to B (1Eꢀ6 cm/s)3 (efflux ratio)
Plasma protein binding % free (mouse, rat, dog, human)
Hepatocyte Clint (ꢀl/min/10ꢀ6 cells) (mouse, rat, dog, human)
hERG IC50 (ꢀM)
62.3 (2.6)
4.9, 2.2, 6.8, 3.5
5, 17, <1, <1
>33.3
Mouse4
7, 0.7, 2, 100
8, 0.6, 3, 79
2, 0.3, 4, 69
Pharmacokinetics
Rat5
Cl(ml/min/kg), Vdss (l/kg), t1/2 (h), F%
Dog6
1Measured LogD at pH7.4.2Solubility of crystalline Form A measured in pH6.5 phosphate buffer at 37°C
3Permeability measured at pH6.5 using 10 ꢀM solution of compound. PK in CD1 Mouse following oral (21
4
5
ꢀMol/kg) and iv doses (10 ꢀMol/kg). PK in Han Wistar Rat following oral (2.1 ꢀMol/kg) and iv doses (1
ꢀMol/kg). 6PK in Dog following oral (2.1 ꢀMol/kg) and iv doses (1 ꢀMol/kg). Half lives reported are from oral
dosing.
ACS Paragon Plus Environment

Page 23 of 39
Journal of Medicinal Chemistry
1
2
3
Figure 5. Coꢀcrystal structure of 44 bound to cꢀKIT. (resolution 2Å). Water molecules shown as red spheres.
4
5
PDB code: 6GQM.
6
7
8
9
Compound 44 was compared against the approved and investigational KIT inhibitors outlined
in Figure 1 against a much broader panel of KIT mutant Ba/F3 cell lines expressing a diverse
set of clinically relevant primary and secondary mutations, including a small panel of PDGFR
driven lines relevant in subsets of GIST (Table 9). It shows very potent growth inhibition of
nearly all mutants tested and consistent with its lineage and other KIT inhibitors, maintains
potent activity against PDGFR, including the clinically GISTꢀrelevant D842V mutant. It
maintains low nM potency against secondary mutations on a primary background of either
exon 9 insertion or exon 11 deletion and is subꢀ10 nM in 12 of the 19 lines examined.
Conversely, imatinib 1 is a relatively poor inhibitor of these diverse KIT mutants, showing
significant activity against a primary mutation of exon 11 insertion and V560D, together with
PDGFR only. Sunitinib 2 carries greater KIT mutant activity than 1 and shows potent activity
against secondary mutations of ATP binding pocket, but is much weaker against clinically
relevant mutations of the Aꢀloop. Third line agent regorafenib 3 is a little more potent against
these mutations but against many of the others appears inferior to sunitinib 2. The
investigational agent ripretinib 4 does show potent and broadꢀspectrum activity against the
KIT mutants in this panel, although with the exception of its PDGF activity it is only superior
to 44 in a single mutant (primary D816V). As previously indicated, the margin to KDR for
this agent is significantly eroded relative to 44 (only 8ꢀfold compared with 153ꢀfold against
exon 11 insertion + D816H). Avapritinib 5 is potent against certain subsets of KIT mutants,
notably D816/820 and PDGFR, although lacks potent activity against mutants such as V654A
and T670I. Consistent with our goals, compound 44 demonstrates equivalent or superior
growth inhibition in all but 2 of the 19 cell lines investigated when compared to agents 1 to 5.
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
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 39
1
2
3
4
5
6
7
Table 9. Comparison of mutant KIT and PDGF potencies for compound 44 with approved and investigational
GIST therapies 1 ꢀ 5.
8
9
KIT inhibitor Ba/F3 GI50 (ꢀM)
Primary
mutation
Secondary
Mutation
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
Gene
44
1
2
3
4
5
Exon 9 ins1
Exon 9 ins1
Exon 9 ins1
Exon 11 del2
Exon 11 del2
Exon 11 del3
Exon 11 del2
Exon 11 del2
Exon 11 del2
Exon 11 del2
Exon 11 del2
Exon 11 del2
V560D
ꢀ
0.002
0.022
0.051
0.001
0.003
0.008
0.016
0.009
0.003
0.014
0.003
0.004
0.001
0.007
0.018
0.971
0.001
0.008
0.022
0.192
3.947
2.749
0.017
0.392
1.523
9.689
0.537
0.216
0.669
0.223
0.312
0.070
0.701
1.191
9.930
0.046
0.102
0.567
0.007
0.012
0.584
0.004
0.006
0.024
0.005
0.400
0.214
0.782
0.271
0.430
0.027
0.007
0.696
0.638
0.030
0.114
0.114
1.270
0.833
0.021
0.234
0.943
0.034
0.297
0.063
0.094
0.049
0.047
0.108
0.549
0.834
2.371
0.051
0.029
0.522
0.034
0.205
0.078
0.004
0.032
0.130
0.038
0.013
0.006
0.014
0.004
0.003
0.023
0.075
0.048
0.037
0.040
0.045
0.188
0.167
0.751
0.069
0.078
0.292
0.117
0.860
0.017
0.019
V654A
D816H
ꢀ
V654A
V654A
T670I
D816H
D820A
Y823D
N822K
A829P
ꢀ
KIT
0.034
0.028
0.042
0.427
0.029
0.008
0.034
0.022
0.010
V560D
V654A
D816H
ꢀ
V560D
D816V
Telꢀ
PDGFRα
Telꢀ
PDGFRβ
V561D4
PDGFR
D842V4
0.631
1
2
3
4
AY502ꢀ503 insertion at exon 9. Deletion of 557ꢀ558 at exon 11. Deletion of 560ꢀ578 at exon 11.
PDGFRα
In order to ensure that the broad KIT mutant inhibition profile observed had not been
achieved at the expense of general kinome selectivity, compound 44 was assessed in an
ACS Paragon Plus Environment

Page 25 of 39
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
extended panel of human kinase binding assays alongside clinical agents 1 – 5 (Figure 6). In
this panel of 379 human kinases, activity was compared by screening at a single, relatively
high concentration of 1 ꢀM, which is far in excess of the potency seen in the Ba/F3 GI50
assays. Encouragingly compound 44 compares well to these other inhibitors, with an overall
selectivity similar to that of imatinib 1 and avapritinib 5. Both Sunitinib 2 and regorafenib 3
show increased activity across this panel, with exploratory agent ripretinib 4 showing a
spectrum of activity between these two approved therapies.
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
Figure 6. Comparative selectivity of inhibitors 1 – 5 and 44 in a panel of 379 human kinases run in the
SelectScreen kinase panel at ThermoFisher Scientific at a single concentration of 1 ꢀM. This data is available as
individual annotated kinase datapoints in Supporting Information.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 39
1
2
3
4
5
6
7
8
9
A key advantage of using Ba/F3 cell lines in the cascade optimisation is that these allow a
direct assessment of potency translation in vivo. The activity of compound 44 in a series of
mouse allograft tumor models using these Ba/F3 lines was assessed and compared to current
standards of care relevant to each mutant (Figure 7 and Table 10). As expected 1 shows only
limited activity in the exon 11 del/D816H background a mutation to which it is refractory.
Regorafenib 3 fares better and is able to drive regression of tumour volume at a dose of 100
mg/kg qd, although this is surpassed by 44 at a dose of 20 mg/kg bid. A dose response is
evident in that 44 at a dose of 2 mg/kg bid shows no impact on tumour size. Sunitinib 2 was
not assessed in this model as it is inactive against the D816H mutant. In an exon 11
del/V654A allograft however and consistent with its clinical use, 2 showed strong regressions
at 80 mg/kg qd, and was matched by 44 at a dose of 20 mg/kg bid (and even as low as 2
mg/kg 44 was highly effective). Imatinib showed moderate activity in this study. Figure 8
shows the pharmacodynamic response to 44 in the D816H model at the top 20 mg/kg dose
tested with the impact on levels of pKIT and downstream markers pERK and pAKT, together
with a timecourse of compound free cover levels above IC90 in the same study. Following a
single dose, high free cover over target is achieved and is associated with complete
suppression of pKIT and downstream markers out to 8 hours. As compound is cleared and
cover declines, these signals recover although recovery is not complete even at 24 hours.
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
ACS Paragon Plus Environment

Page 27 of 39
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
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
Figure 7. The in vivo efficacy of KIT inhibitors in mouse allograft disease models. Upper panel: Imatinib and
regorafenib versus 44 in an exon 11 del/D816H model. Lower panel: Imatinib and sunitinib versus 44 in an exon
11 del/V654A model. Doses for standards of care were selected to best represent their clinically relevant doses.
Table 10. Percent and statistical significance of tumor growth inhibition and regressions of 44 relative to vehicle
and standards of care in two mouse allograft disease models of mutant KIT activity.
Dose
(mg/kg)
%
%
p-
Tumor
ID
vehicle
1
Schedule Route
Inhibition1
value2
Regression1
n/a
BD
QD
p.o
p.o
n/a
43
n/a
0
n/a
Ba/F3 KITꢀexon 11
del/D816H
300
0.0095
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 39
1
2
3
4
5
6
7
8
9
100
2
QD
BD
BD
BD
QD
QD
BD
BD
p.o
p.o
p.o
p.o
p.o
p.o
p.o
p.o
>100
5
39
0
<0.0001
0.3818
<0.0001
n/a
3
44
20
n/a
300
80
2
>100
n/a
75
n/a
0
44
vehicle
1
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
34
0.0495
<0.0001
<0.0001
<0.0001
Ba/F3 KITꢀexon 11
del/V654A
>100
>100
>100
87
44
85
2
44
20
44
1 On last day of dosing. 2 Relative to vehicle.
Figure 8. Pharmacodynamic effect of 44 on tumor pKIT, pERK and pAKT in an exon 11 del/D816H mouse
allograft model from a single dose of 20 mg/kg po. No pKIT was detectable at t=2 or 8 hours. A more detailed
statistical analysis of this data is available in Supporting Information.
Conclusions
Having established as a goal the challenging profile of an inhibitor capable of addressing all
diverse KIT mutants, this has been realised in a series of type II quinazoline kinase inhibitors.
Identification of a series of kinase inhibitors from an historical project, combined with
structureꢀbased insights allowed for rapid optimisation in particular of both mutant and kinase
ACS Paragon Plus Environment

Page 29 of 39
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
selectivity, not just against KDR but the wider kinome. Despite high homology between KIT
and antiꢀtarget KDR, we propose that subtle differences in the way each is able to
accommodate and in particular solvate the liganded structures is key to the enhanced
selectivity seen. The preꢀclinical in vivo efficacy of 44, now designated AZD3229 is
encouraging, particularly in relation to current standards of care and may hold promise in the
future treatment of GIST patients. Further data relating to the pharmacological activity of
AZD3229, including assessment of required margins over KDR to ensure no hypertensive
effects31 that limit current agents will be reported in due course.
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 Methods
Chemistry Unless otherwise stated, commercially available reagents were used as supplied.
All reactions requiring anhydrous conditions were conducted in dried apparatus under an
atmosphere of nitrogen. Flash column chromatography was performed on Merck Kieselgel
silica (Art. 9385) or on reversed phase silica (Fluka silica gel 90 C18) or on Silicycle
cartridges (40ꢀ63 ꢁm silica, 4 to 330 g weight) or on Grace resolv cartridges (4 – 120 g) or on
RediSep Rf 1.5 Flash columns or on RediSep Rf high performance Gold Flash columns (150
– 415 g weight) or on RediSep Rf Gold C18 Reversedꢀphase columns (20 – 40 ꢁm silica)
either manually or automated using an Isco CombiFlash Companion system or similar
system. Preparative reverse phase HPLC was performed on a Waters instrument (600/2700 or
2525) fitted with a ZMD or ZQ ESCi mass spectrometers and a Waters XꢀTerra or a Waters
XꢀBridge or a Waters SunFire reverseꢀphase column (Cꢀ18, 5 microns silica, 19 mm or 50
mm diameter, 100 mm length, flow rate of 40 ml / minute) using decreasingly polar mixtures
of water (containing 1% ammonia) and acetonitrile or decreasingly polar mixtures of water
(containing 0.1% formic acid) and acetonitrile as eluents. NMR chemical shift values were
ACS Paragon Plus Environment