Discovery of Pemigatinib: A Potent and Selective Fibroblast Growth
Factor Receptor (FGFR) Inhibitor
Liangxing Wu,* Colin Zhang, Chunhong He, Dingquan Qian, Liang Lu, Yaping Sun, Meizhong Xu,
Jincong Zhuo, Phillip C. C. Liu, Ronald Klabe, Richard Wynn, Maryanne Covington, Karen Gallagher,
Lynn Leffet, Kevin Bowman, Sharon Diamond, Holly Koblish, Yue Zhang, Maxim Soloviev,
Gregory Hollis, Timothy C. Burn, Peggy Scherle, Swamy Yeleswaram, Reid Huber, and Wenqing Yao
Cite This: https://doi.org/10.1021/acs.jmedchem.1c00713 Read Online
ACCESS Metrics & More Article Recommendations *sı Supporting Information
ABSTRACT: Aberrant activation of FGFR has been linked to the pathogenesis of
many tumor types. Selective inhibition of FGFR has emerged as a promising approach
for cancer treatment. Herein, we describe the discovery of compound 38
(INCB054828, pemigatinib), a highly potent and selective inhibitor of FGFR1,
FGFR2, and FGFR3 with excellent physiochemical properties and pharmacokinetic
profiles. Pemigatinib has received accelerated approval from the U.S. Food and Drug
Administration for the treatment of adults with previously treated, unresectable locally
advanced or metastatic cholangiocarcinoma with a FGFR2 fusion or other
rearrangement. Additional clinical trials are ongoing to evaluate pemigatinib in
patients with FGFR alterations.
■ INTRODUCTION
Fibroblast growth factor receptors (FGFRs) are a family of
receptor tyrosine kinases with an extracellular ligand binding
domain, a transmembrane domain, and a cytosolic tyrosine
kinase domain. Binding of fibroblast growth factor (FGF)
ligands to the extracellular domain of FGFRs causes
dimerization of the receptor, which then activates the
intracellular tyrosine kinase domain. This leads to receptor
autophosphorylation and activation of multiple downstream
signaling pathways including RAS-MAPK, PI3K-AKT, STAT,
and PLCγ. FGFR/FGF signaling plays an essential role in cell
proliferation, survival, and motility. FGFRs are composed of
four highly conserved members (FGFR1, FGFR2, FGFR3, and
FGFR4) that are expressed in a variety of cells.1,2
Genetic alterations such as gene amplification, point
mutation or chromosomal translocation/fusion can result in
constitutive activation of FGFRs or abnormal ligand-dependent signaling, which can potentially lead to tumor formation.
There is strong evidence that dysregulation of FGFRs is
involved in diverse tumor types.3−6 For example, chromosomal
translocation of FGFR1 at the 8p11 locus with partner genes
has been reported to be responsible for 8p11 myeloproliferative syndrome.7 Similarly, fusion of FGFR2 with other genes
is reported in 10−15% of intrahepatic cholangiocarcinoma
cases.8 Activating point mutations in FGFR3 have been found
to be enriched in bladder cancer and could be one of the
potential driving oncogenes.9 Furthermore, FGFR110 and
FGFR211 amplification are found to be prevalent in squamous
cell lung cancer and gastric cancer, respectively, and are
potentially associated with tumor growth.
Due to the prevalence of abnormal FGFR activity across a
variety of cancer types, FGFR inhibition represents an
attractive therapeutic approach for the treatment of cancers
with genetic FGFR alterations.12 Several small molecule FGFR
inhibitors such as dovitinib13 and brivanib alaninate14 (Figure
1) have progressed into clinical studies and demonstrated
promising activities.6 However, these first generation FGFR
inhibitors also potently inhibited a broad range of other kinases
including VEGFR2. The lack of selectivity may cause doselimiting toxicity, thereby potentially reducing the therapeutic
window for FGFR-driven cancers with these early inhibitors.12
Small molecules that selectively inhibit FGFRs could offer
favorable toxicity profile and improved therapeutic window.
Therefore, we set out to discover potent and highly selective
FGFR inhibitors for the treatment of tumors with FGFR
alterations. At the time we initiated our discovery campaign,
little was known on how to develop selective small molecule
Received: April 19, 2021
pubs.acs.org/jmc Drug Annotation
© XXXX American Chemical Society A
https://doi.org/10.1021/acs.jmedchem.1c00713
J. Med. Chem. XXXX, XXX, XXX−XXX
Downloaded via KUNMING UNIV OF SCIENCE AND TECHNOLOGY on August 10, 2021 at 20:36:00 (UTC).
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
FGFR inhibitors due to the high structural similarity of FGFRs
with several other kinases, such as VEGFR2. A literature review
of FGFR X-ray crystal structures and reported FGFR inhibitors
revealed a tool compound PD173074 (Figure 1), which
appeared to exhibit decent FGFR potency with reasonable
selectivity over other kinases including VEGFR2.15,16 After
studying the cocrystal structure of PD173074 with FGFR1, we
reasoned that the high affinity and selectivity for FGFR1 might
result from the filling of a complementary hydrophobic pocket
near the FGFR gatekeeper region with a 3,5-dimethoxyphenyl
substituent. These findings provided a good starting point for
the development of second-generation FGFR selective
inhibitors with representative examples shown in Figure 2.
17
Herein, we report our discovery efforts including structurebased drug design, structure−activity relationship (SAR)
studies, and improvement of physiochemical properties and
pharmacokinetics (PK) that culminated in the identification of
the clinical candidate compound 38 (INCB054828, pemigatinib), a potent and selective inhibitor of FGFR1, FGFR2 and
FGFR3.
Figure 1. Structures of representative FGFR inhibitors.
Figure 2. Structures of representative pan-FGFR selective inhibitors in clinical development.
Figure 3. Chemical structure of screening hit 1 and subsequent design of a tricyclic series as potent and selective FGFR inhibitors.
Journal of Medicinal Chemistry pubs.acs.org/jmc Drug Annotation
https://doi.org/10.1021/acs.jmedchem.1c00713
J. Med. Chem. XXXX, XXX, XXX−XXX
B
■ RESULTS AND DISCUSSION
To identify a novel starting point for potent and selective
FGFR inhibitors, we first conducted a high-throughput screen
of our in-house compound library (a focused library of about
20 000 compounds from in-house historical projects) using a
HTRF enzymatic assay with the FGFR1 enzyme. Among the
various hits from the screening campaign, compound 1 (Figure
3) showed strong inhibition of FGFR1 (IC50 = 6 nM) with
good selectivity over VEGFR2 (IC50 = 2058 nM) and seemed
to be a promising starting point. We hypothesized that the
pyrrolopyrimidine moiety in compound 1 might bind to the
hinge of FGFR1 through two hydrogen bonds and offered an
intriguing scaffold for further exploration. While compound 1
was a potent inhibitor for FGFR1, it also potently inhibited
JAK2 (IC50 < 1 nM) and several other kinases. Preliminary
SAR indicated that a major challenge for this pyrrolopyrimidine series would be improving general kinase selectivity.
We reasoned that introduction of a 3,5-dimethoxyphenyl
group into the pyrrolopyrimidine scaffold of compound 1
could potentially improve kinase selectivity. After examining
the superimposition of pyrrolopyrimidine 1 with PD173074,
we hypothesized that the pyrimidine nitrogen of compound 1
could be a good position to attach the 3,5-dimethoxyphenyl
substituent. To test this hypothesis, we prepared a hybrid
compound 2 (Figure 3), which was found to significantly
improve selectivity over JAK2 and maintained decent
selectivity over VEGFR2, albeit with a loss in FGFR1 potency
relative to 1. After careful conformational analysis of
compound 2, we hypothesized that FGFR activity might be
restored by cyclizing the ethylene linker onto the pyrrolopyridine core to rigidify the structure and properly project the 3,5-
dimethoxyphenyl group into the hydrophobic pocket near the
FGFR gatekeeper region. Thus, tricyclic compound 3 was
prepared, which featured a polar cyclic urea linker with some
sp3 character that we envisioned could be beneficial for future
improvement of physiochemical properties.18 Indeed, compound 3 showed promising FGFR1 potency while maintaining
good selectivity over JAK2, and more importantly it also
exhibited good selectivity over the closely related VEGFR2.
Having demonstrated encouraging biochemical potency and
selectivity, compound 3 was selected as a promising starting
point for further investigation.
To guide SAR studies, the following in vitro assays were
developed: (1) biochemical enzymatic assays for FGFR1,
FGFR2, FGFR3, and VEGFR2 to gauge selectivity; (2)
proliferation assays using FGFR1-dependent H1581 cell line
and FGFR2-dependent KatoIII cell line; and (3) a whole
blood (WB) assay to estimate compound potency in an in vivo
biological system, measuring inhibition of pFGFR2 with
KatoIII cells spiked into human whole blood. Compound 3
was found to inhibit FGFR1, FGFR2 and FGFR3 with good
enzymatic activities (Table 1). It was further evaluated in
cellular assays and demonstrated moderate activity. In an effort
to further improve potency, we targeted SAR exploration in the
gatekeeper region to maximize the interactions in the
hydrophobic pocket by modification of the 3,5-dimethoxyphenyl group. Introduction of a lipophilic chlorine atom at
the ortho-position of the dimethoxyphenyl ring gave compound
4 (Table 1) and resulted in 3-fold improvement in potency
while maintaining good selectivity over VEGFR2. Compound
4 also demonstrated improved cellular activities with
encouraging WB activity (IC50 = 81 nM). Addition of a
second ortho-chlorine atom (compound 5) led to further
improvements in potency as well as VEGFR2 selectivity. In
contrast, addition of a second chlorine atom at the paraposition (compound 6) caused a significant loss in potency.
Mono-ortho-fluoro substitution of the 3,5-dimethoxyphenyl
ring (compound 7) gave comparable potency to the
corresponding chloro analogue (4). Interestingly, introduction
of bis-ortho-fluorine (compound 8) dramatically increased
FGFR potency. Compound 8 demonstrated sub-nanomolar
activity in enzymatic and cellular assays with good selectivity
over VEGFR2. We were also excited that compound 8 showed
single-digit nanomolar activity against KatoIII cells in the WB
assay. The combination of ortho-fluoro and -chloro substitution provided a moderately less potent compound 9,
which displayed a 3-fold loss in WB potency compared to the
difluoro analogue (8). The source for this WB potency loss
may be related to the difference in plasma protein binding for
these two compounds (2.4% and 4.7% free for 9 and 8,
respectively).
Table 1. SAR Investigation of the 3,5-Dimethoxylphenyl Substituent to Improve Potency
enzyme IC50 (nM) cellular IC50 (nM) WB IC50 (nM)
R1 R2 R3 FGFR1 FGFR2 FGFR3 VEGFR2 H1581 KatoIII KatoIII
3 H H H 7.5 4.4 5.7 452 40 42 −a
4 Cl H H 1.9 1.4 2.2 131 11 4.3 81
5 Cl Cl H 1.1 1.0 1.2 162 4.7 1.8 65
6 Cl H Cl 82 97 108 >2000 538 356 −a
7 F H H 1.1 1.4 1.4 92 5.5 4.7 67
8 F F H 0.12 0.15 0.27 24 1.3 0.86 6
9 F Cl H 0.17 0.3 0.49 40 2.1 1.3 18
−: not tested.
Journal of Medicinal Chemistry pubs.acs.org/jmc Drug Annotation
https://doi.org/10.1021/acs.jmedchem.1c00713
J. Med. Chem. XXXX, XXX, XXX−XXX
C
Having achieved good FGFR potency and selectivity over
VEGFR2, additional characterization was carried out for
compound 8. The general kinase selectivity was evaluated
internally in a 56-kinase panel, and 8 demonstrated good
selectivity with only three kinases displaying IC50 values less
than 1 μM (c-KIT: 12 nM; PDGFRβ: 406 nM; and TRKA:
458 nM). The in vitro absorption, distribution, metabolism,
and excretion (ADME) properties of compound 8 were also
encouraging. We observed good permeability in the Caco-2
assay (6.4 × 10−6 cm/s) and low intrinsic clearance in human
liver microsomes testing (H-Cl: 0.5 L/h/kg) with no hERG
liability for compound 8. On the basis of these results,
compound 8 was advanced into in vivo PK studies. When
administered intravenously (IV: 1 mg/kg) to rats in a cassette
study, compound 8 exhibited low clearance (10% of hepatic
blood flow (HBF)), a moderate volume of distribution (Vdss:
0.8 L/kg), and a decent half-life (t1/2) of 2.8 h. At a dose of 2
mg/kg, excellent oral exposure was observed with an AUC of
10 600 nM·h and 58% oral bioavailability. Despite the excellent
rat PK profile, compound 8 performed relatively poor in
cynomolgus monkey (cyno) PK with several potential issues:
(1) a low oral exposure of 1690 nM·h at 2 mg/kg dose and
poor bioavailability (15%); (2) a short t1/2 of 1.1 h, indicating
that the compound would require a more frequent dosing
regimen to maintain desired exposure and avoid high Cmax
driven side effects;19,20 and (3) a small Vdss of 0.3 L/kg, which
was conceived to give rise to poor tissue penetration and
limited in vivo efficacy.21 The chloro analogue 9 was also
evaluated in PK studies (Tables S5 and S6) and demonstrated
very similar PK profiles as compound 8.
Given that compound 8 demonstrated good Caco-2
permeability and low clearance in cyno PK, we hypothesized
that the poor exposure and low oral bioavailability may be the
consequence of low aqueous solubility.22,23 Upon further
investigation, we discovered compound 8 indeed exhibited
poor solubility (0.3 μg/mL) in fasted state simulated intestinal
fluid (FaSSIF). To improve upon compound 8, we focused our
efforts on the addition of polar functionality to the N1 position
(R4
) of the cyclic urea. We proposed that addition of polarity
from this vector may lead to increased solubility and reduced
lipophilicity and could potentially also lower clearance to
extend t1/2.
24,25 A variety of polar groups appended to the
cyclic urea were found to be well tolerated with only a minor
effect on potency. Addition of an ethyl-linked hydroxyl group
(10, Table 2) increased solubility by 10-fold while also
maintaining good Caco-2 permeability (5.1 × 10−6 cm/s)
albeit slightly higher in vitro clearance (0.9 L/h/kg). However,
10 exhibited worse rat PK compared to 8, with a shorter t1/2
and lower exposure. While typically only the compounds with
good rat PK properties would be advanced into cyno PK
studies, 10 was progressed to investigate possible improvements resulting from improved aqueous solubility. Unfortunately, compound 10 showed very poor cyno PK properties
with low exposure and oral absorption. Intrinsic clearance was
slightly lowered (0.8 L/h/kg) by rigidifying the structure to the
cyclopentyl alcohol (11), which also gave another 10-fold
improvement in FaSSIF solubility. Disappointingly such
modification led to even worse rat PK profile. Basic amine
groups such as ethylene linked methylpiperazine (12) were
also examined.26 Compound 12 showed significantly improved
Table 2. Modification of Urea Substitution to Improve Solubility and PK Properties
a
Dosed in cassette studies, 1 mg/kg IV and 2 mg/kg PO. b
0.5 mg/kg IV and 2 mg/kg PO; “−” not tested.
Journal of Medicinal Chemistry pubs.acs.org/jmc Drug Annotation
https://doi.org/10.1021/acs.jmedchem.1c00713
J. Med. Chem. XXXX, XXX, XXX−XXX
D
solubility and decent in vitro ADME properties. When
advanced to a rat PK study, we were surprised again to
observe very poor exposure and oral bioavailability. Compound 13 featured reduced flexibility by directly linking a
piperidine group to the cyclic urea. This change resulted in
slightly decreased intrinsic clearance but ultimately provided a
poor rat PK profile with low exposure and oral bioavailability.
Extensive SAR studies with other polar groups aimed to
improve solubility also did not translate to improvement in PK
properties.
A second strategy we explored was the incorporation of
nonpolar groups to potentially decrease the planarity of the
molecule and disrupt crystal packing thereby increasing
solubility.27 Addition of a cyclopropyl group (compound 14)
was found to increase the solubility (11 μg/mL) relative to
compound 8 while also maintaining good rat PK properties.
Compound 14 was advanced to cyno PK, where we observed a
poor profile as compared to compound 8. Further increasing
lipophilicity with other nonpolar groups was found to be
detrimental for PK properties. The improvement of physiochemical properties from the R4 vector did not translate into
positive PK results, therefore we redirected our focus to other
regions of the molecular scaffold.
We hypothesized that a potential cause for the repeatedly
poor cyno PK could be the electron-rich pyrrole ring in the
scaffold, which could be metabolically labile and susceptible to
oxidation.28 We proposed that replacing the pyrrole ring with a
pyrazole might not only stabilize the molecule but the
additional polar nitrogen may also improve solubility. Indeed,
the pyrazole analogue 15 demonstrated improved solubility of
4 μg/mL in FaSSIF while maintaining decent potency and a
good ADME profile (Table 3). Compound 15 showed an
excellent rat PK profile but failed to provide any significant
improvement in cyno PK relative to the pyrrole analogue 8.
Conversion of the pyrrole ring to imidazole (compound 16)
led to further improvement in FaSSIF solubility (21 μg/mL).
Compound 16 had good in vitro ADME properties and
excellent rat PK profile but, disappointingly, inferior exposure
in cyno PK compared to the pyrrole compound 8.
Serendipitously, we discovered that the pyrrole ring could be
oxidized to the corresponding lactam while trying to introduce
a fluoro substitution using Selectfluor. After improving the
reaction conditions, we were able to efficiently convert the
pyrrole of compound 8 to the lactam (17) in the presence of
pyridinium tribromide followed by reduction with zinc in high
yield. We were pleased to find that 17 maintained decent
potency with excellent in vitro ADME properties and rat PK
profile. Encouragingly, compound 17 had slightly improved
cyno PK properties compared to the pyrrole precursor 8 with a
longer t1/2 (2.8 h), higher exposure (2109 nM·h) and better
bioavailability (24%). With the lactam core, we quickly
revisited substitution on the cyclic urea. Replacement of the
N-methyl in 17 with an ethyl group provided compound 18
and led to nearly a 2-fold improvement in WB potency. We
were also thrilled to discover that this modification led to
significant improvement in cyno PK exposure (12 122 nM·h at
2 mg/kg dose) with high oral bioavailability (73%). However,
the improvement in PK properties by ethyl substitution
appeared to be preferable for the lactam series, as converting
compound 15 to the ethyl derivative 19 provided no significant
improvement in PK profiles.
Given its good potency and excellent PK profiles, compound
18 became our compound of interest for further in vitro and in
vivo evaluations. In enzymatic assays, 18 displayed strong
FGFR1, FGFR2, and FGFR3 inhibition activity with IC50 of
1.3 nM, 2 nM, and 2.7 nM, respectively. In the cell
proliferation assays, IC50 values of 17 nM (H1581) and 4.8
nM (KatoIII) were observed. Additionally, this compound
exhibited low hERG patch clamp activity (0% inhibition at 5
μM) and was not a potent inhibitor of cytochrome P450
(CYP) with IC50 values >25 μM for the major CYP isozymes
including CYP1A2, CYP2C9, CYP2C19, CYP2D6, and
CYP3A4. Kinase profiling with an in-house 56-kinase panel
revealed that compound 18 was highly selective, inhibiting just
two other kinases with IC50 less than 1 μM (VEGFR2: 288 nM
and c-KIT: 407 nM). Data supported compound 18 to be
Table 3. Scaffold Hopping to Improve PK Properties
CL (HBF%)/Vdss (L/kg)/t1/2 (h)/AUC (nM·h)/
F%
X1
−X2 R4 WB_ KatoIII
IC50(nM)
Sol in FaSSIF
(μg/mL)
Caco-2
(×10−6
cm/s)
H-Cl
(L/h/kg) rat PK profilea cyno PK profilea
15 CHN Me 32 4 8.1 0.5 27/0.64/0.8/8300/
>100
21/0.40/0.8/1775/21b
16 NCH Me 22 21 6.8 0.6 29/0.99/1.9/10800/
>100
29/0.27/0.9/442/6
17 CH2CO Me 35 10 3.4 <0.5 34/1.33/1.3/3500/74 12/0.68/2.8/2109/24b
18 CH2CO Et 20 10 5.7 0.6 11/0.51/1.1/27300/− 11/0.46/2.1/12122/
73b
19 CHN Et 18 3 8.3 0.9 20/0.47/0.7/8000/
>100
12/0.25/1.6/2635/16
a
Dosed in cassette studies, 1 mg/kg IV and 2 mg/kg PO. b
0.5 mg/kg IV and 2 mg/kg PO.
Journal of Medicinal Chemistry pubs.acs.org/jmc Drug Annotation
https://doi.org/10.1021/acs.jmedchem.1c00713
J. Med. Chem. XXXX, XXX, XXX−XXX
E
scaled up for in vivo pharmacology and toxicology studies.
During the scale up, crystalline material was obtained that was
later shown to give significantly lower oral exposure and
bioavailability in both rat and cyno (5 mg/kg oral dosing in rat:
AUC 8700 nM·h, F% = 17; 5 mg/kg oral dosing in cyno: AUC
= 6410 nM·h, F% = 17) compared to the earlier studies
performed with amorphous material. Moreover, dose escalation studies in rat with the crystalline material only achieved a
2.8-fold increase in AUC when the dose was increased by 10-
fold to 50 mg/kg. These observations would impair the ability
to achieve suitable exposures for future studies. We reasoned
that the limited exposure could potentially be associated with
reduced solubility for the crystalline material (1 μg/mL in
FaSSIF, compared to 10 μg/mL for the amorphous material).
Extensive formulation studies including addition of surfactant,
nanosuspension, or nonaqueous formulation with polyethylene
glycol did not provide any significant improvement in
exposure.29 Properties of the crystalline material hampered
further development of compound 18. We subsequently
prioritized our SAR efforts to identify lactam analogues with
improved aqueous solubility.
Given our experiences on the pyrrole-containing scaffold, we
focused our initial effort to improve aqueous solubility on the
addition of polarity to the R4 position of the cyclic urea.
However, we quickly discovered that polar groups were not
tolerated in this region for the lactam series (Table 4). For
example, addition of a hydroxyl (20) or basic morpholine (21)
to the ethyl group in compound 18 caused significant loss of
activity in both the enzymatic and cellular assays. A variety of
other polar groups were also found to be detrimental for
potency. We then investigated whether the addition of
lipophilic substitution could increase aqueous solubility while
preserving activity. After systematic SAR studies, 2,3-
difluorophenyl derivative (22) was identified as a promising
compound with good potency and improved aqueous
solubility. We reasoned that the 2,3-difluorophenyl group
could potentially disrupt molecular symmetry and planarity,
thereby weakening crystal packing and leading to improved
solubility.27 More encouragingly, compound 22 maintained a
reasonable cyno PK profile with low clearance (7% HBF) and
decent half-life (t1/2) of 3.7 h, albeit a small volume of
distribution (Vdss: 0.26 L/kg), in a 0.5 mg/kg IV dosing study.
Good oral exposure was observed with an AUC of 5529 nM·h
at a dose of 2 mg/kg and a moderate oral bioavailability (21%).
Disappointingly, 22 was later found to display time-dependent
inhibition (TDI) of human CYP3A4. Specifically, preincubation of 20 μM of compound 22 with human CYP3A4 for 30
min led to a 30% loss of the enzyme activity. Given that TDI of
human CYP3A4 indicated the potential formation of reactive
metabolites that may lead to toxicity or cause drug−drug
interactions (DDI) in patients, we next focused on mitigating
the TDI issue.30 The ethyl analogue 18 had minimal effect of
CYP3A4 activity under the same conditions, suggesting that
the 2,3-difluorophenyl moiety in compound 22 might be
responsible for the observed TDI. Therefore, our efforts were
focused on modification of the 2,3-difluorophenyl group. We
did not observe a clear SAR trend for TDI except that reducing
lipophilicity seemed to alleviate the issue. This presented a
significant challenge for the lactam series as we had observed
with our earlier SAR that polar groups in the R4 region were
found to be detrimental for potency. For example, the addition
of a para-cyano group to the phenyl ring (compound 23)
eliminated the TDI activity but led to a loss of potency.
Extensive SAR efforts in this region failed to mitigate the TDI
liability while preserving other properties.
Returning our focus to the goal of enhancing solubility while
maintaining good activity, we proposed modification of the α-
carbon of the lactam carbonyl with spirocyclic or geminal
substitution as a means to decrease molecular planarity and
increase sp3
-character. Attachment of a spirocyclopropyl ring
(compound 24, Table 4) led to significant decrease in FGFR
potency. Interestingly, converting the ethyl in 24 to a methyl
group (25) recovered the potency; however, compound 25
exhibited very poor aqueous solubility (0.3 μg/mL in FaSSIF).
Opening of the cyclopropyl ring to the gem-dimethyl analogue
26 decreased the potency precipitously. Attempts to enhance
aqueous solubility via addition of polar substituents, such as
spiropiperidine (compound 27), were also found to be
detrimental for potency. Although the lactam series offered
several promising compounds, we were unable to discover a
molecule with the desired balance of potency, solubility,
ADME properties, and PK profiles.
At this point, we decided to revisit the pyrrole series,
hypothesizing that introduction of substitution on the pyrrole
moiety to fine-tune the steric and electronic properties may
reduce the structural liability and improve PK profiles.28
Furthermore, these modifications may also help to improve
aqueous solubility. Functionalization of the 3-position (R7
) of
the pyrrole ring with a pyrazole substituent (28, Table 5) was
found to mostly preserve potency and increase solubility
compared to the parent compound 8. Unfortunately, 28
exhibited potent inhibition of CYP2C9 with an IC50 of 2 μM.
Modification of R7 did not lead to an appropriately balanced
Table 4. SAR to Improve Solubility for the Lactam Series
Journal of Medicinal Chemistry pubs.acs.org/jmc Drug Annotation
https://doi.org/10.1021/acs.jmedchem.1c00713
J. Med. Chem. XXXX, XXX, XXX−XXX
F
profile of potency, ADME properties, CYP inhibition, and
solubility.
We redirected our focus to the 2-position (R8
) of the pyrrole
ring, which would project substitution toward the solventexposed region. We proposed that this pocket could
accommodate various structural modifications to improve
ADME and PK properties. Indeed, many substituents at R8
were found to be well tolerated for potency. Aromatic
substitution at this position led to very potent compounds.
For example, addition of a pyrazole ring gave compound 29
that displayed cellular IC50 values of 0.06 and 0.7 nM in the
H1581 and KatoIII proliferation assays, respectively. However,
this modification significantly reduced general kinase selectivity
such that 29 potently inhibited a variety of off-target kinases
including VEGFR2 (IC50 = 1.3 nM), TRKA (IC50 < 1 nM),
AURKB (IC50 = 2.5 nM), c-KIT (IC50 = 3.9 nM), and
PDGFRβ (IC50 = 5.7 nM). We shifted our attention to
examine if polar amide substituents would be tolerated and
potentially could decrease the electron-density of the pyrrole
ring to reduce perceived metabolic liabilities. Dimethyl amide
derivative 30 maintained FGFR potency, displayed good
general kinase selectivity, and showed slightly improved
solubility (2 μg/mL). Compound 30 presented excellent rat
PK with high oral exposure and good bioavailability; however,
30 showed no measurable oral exposure and bioavailability in
cyno PK. Functionalization of the amide with piperazine (31)
led to enhanced solubility. While 31 showed improved cyno
PK properties with slightly better exposure and bioavailability,
we were disappointed that further modification of the amide
did not yield additional improvement. Direct attachment of a
saturated heterocycle such as piperidine (compound 32) to the
pyrrole ring led to decreased potency. Compound 32 still
advanced into a rat PK study due to its good aqueous solubility
(44 μg/mL in FaSSIF) and in vitro ADME properties (Caco-2:
Table 5. SAR Efforts to Improve PK Properties for Pyrrole Series
a
Dosed in cassette studies (IV: 1 mg/kg; PO: 2 mg/kg). b
0.5 mg/kg IV and 2 mg/kg PO. c
Dosed discretely (IV: 1 mg/kg; PO: 2 mg/kg).
Journal of Medicinal Chemistry pubs.acs.org/jmc Drug Annotation
https://doi.org/10.1021/acs.jmedchem.1c00713
J. Med. Chem. XXXX, XXX, XXX−XXX
G
2.9 × 10−6 cm/s; H-Cl: 0.5 L/h/kg). Disappointingly, 32
performed very poorly in rat PK. After numerous SAR studies,
we discovered that the attachment of N-ethylpiperazine via a
C1−3 alkyl linker (compound 33−35) led to excellent aqueous
solubility and decent potency. Interestingly, while the
analogues with 3-carbon linker (33) and 2-carbon linker
(34) demonstrated very poor rat PK profiles, the structurally
similar 1-carbon linker derivative 35 showed a much improved
rat PK. More notably, compound 35 exhibited an excellent oral
exposure of 5873 nM*h at a dose of 2 mg/kg in cyno and
decent oral bioavailability (26%) with a long t1/2 of 12 h in the
IV arm of the study. This represented a major breakthrough in
our search for compounds with acceptable cyno PK profiles in
the pyrrole series. It came as a great disappointment to
discover later that compound 35 showed TDI of CYP3A4
(14% and 45% inhibition at compound concentrations of 10
μM and 25 μM, respectively).
We first hypothesized that the benzylic amine of 35 might be
responsible for the observed TDI.30 Surprisingly, removal of
the benzylic amine (compound 36) led to an undesirable
increase in TDI of CYP3A4 (22% inhibition at 10 μM of 36)
and exhibited very poor rat PK. An alternative cause for the
TDI of compound 35 could be the distal basic nitrogen. We
were pleased to discover that replacement of the nitrogen with
oxygen to give compound 37 seemed to mitigate the TDI issue
with only 4% inhibition of CYP3A4 at 10 μM compound
concentration. Gratifyingly, the morpholine analogue 37 also
afforded much-improved rat and cyno PK profiles (AUC =
8193 and 11804 nM·h in rat and cyno, respectively) compared
to the ethylpiperazine derivative 35, albeit with lower solubility
(17 μg/mL in FaSSIF). Conversion of the methyl substitution
on the cyclic urea to ethyl (compound 38) led to nearly a 3-
fold improvement in WB potency and retained excellent rat
and cyno PK profiles. Compound 38 was also clean of TDI,
indicating that the morpholine replacement had mitigated the
TDI issue. Furthermore, detailed metabolite profiling of
compound 38 did not reveal the formation of reactive
metabolites.
With the new analogues from the pyrrole series, we
successfully resolved the oral exposure and half-life issues in
cyno PK. Nevertheless, there was still some concern that the
Vdss was a little bit low (0.58 L/kg in monkey for compound
38) and could potentially lead to poor in vivo efficacy. During
the course of our SAR studies, extensive efforts were
undertaken to increase Vdss. A variety of strategies including
increasing lipophilicity, introducing basic amines, and increasing plasma free fraction were investigated to improve Vdss.
21 All
endeavors turned out to be counterproductive. Despite its Vdss,
in subsequent studies, compound 38 showed good PK/PD
correlation and achieved good in vivo efficacy. Thus, the lowto-moderate Vdss did not seem to impose a major deterrent to
achieving in vivo activity.
Based on our previous challenges associated with crystalline
material impacting PK properties for the initial lead compound
Scheme 1. Synthesis of Compound 38 (INCB054828, Pemigatinib)a
a
Reagents and conditions: (a) ethylamine, 2-methoxyethanol, 130 °C, 92%; (b) camphorsulfonic acid, xylenes, reflux; (c) LiAlH4, THF, 50 °C,
82% (two steps); (d) triphosgene, THF, RT, 92%; (e) NaH, PhSO2Cl, DMF, 0 °C, 98%; (f) lithium diisopropylamide (LDA), DMF, THF, −78
°C, 91%; (g) morpholine, acetic acid, Na(OAc)3BH, DCM, RT, 95%; (h) tetra-n-butylammonium fluoride (TBAF), THF, 50 °C.
Journal of Medicinal Chemistry pubs.acs.org/jmc Drug Annotation
https://doi.org/10.1021/acs.jmedchem.1c00713
J. Med. Chem. XXXX, XXX, XXX−XXX
H
18, we were interested in evaluating compound 38 as the
crystalline material. During the scale up, crystalline free base of
38 was obtained that still exhibited good solubility especially
under acidic conditions (8 μg/mL in FaSSIF and >6 mg/mL in
simulated gastric fluid (SGF)). More encouragingly, the
crystalline free base gave comparable exposures in rat and
cyno to the earlier studies using the amorphous material, and
exhibited dose-dependent increase in exposures in pharmacology and toxicology studies. On the basis of these results,
compound 38 (INCB054828, pemigatinib) was selected as the
lead candidate of the project to be progressed into preclinical
studies.
The synthesis of compound 38 carried out by the discovery
team is outlined in Scheme 1. A nucleophilic aromatic
substitution reaction replaced the chloride in the commercially
available starting material 39 with ethylamine producing
compound 40. Condensation of aldehyde 40 with 2,6-
difluoro-3,5-dimethoxyaniline 41 formed an imine intermediate, which was then reduced to give the diamine 42. Tricyclic
urea 43 was obtained by cyclization of diamine 42 with
triphosgene. Protection of the pyrrole nitrogen in 43 with a
benzenesulfonyl group (44), followed by selective deprotonation with LDA and quenching with DMF, gave the aldehyde
45. Reductive amination of aldehyde 45 with morpholine,
followed by removal of the sulfonyl protecting group afforded
compound 38 in good yield.
Detailed preclinical characterization of compound 38 was
initially reported in an earlier publication.31 Specifically, 38
potently inhibited FGFR1, FGFR2, and FGFR3 (IC50 = 0.4,
0.5, and 1 nM, respectively) with weaker activity for FGFR4
(IC50 = 30 nM) in enzymatic assays with recombinant human
FGFR kinases. The molecule also demonstrated good cellular
activity and selectively inhibited the growth of FGFR1,
FGFR2, and FGFR3 dependent tumor cell lines with IC50
values in the low nM range (Table S1). Compound 38
exhibited good selectivity over other kinases in our in-house
panel of 56 diverse kinases with only VEGFR2 (IC50 = 71 nM)
and c-KIT (IC50 = 266 nM) being identified with IC50 values
less than 1000 nM (Table S8). Subsequent profiling against a
broader panel of 161 kinases (PerkinElmer, Akron OH) and a
panel of 70 ion channels, receptors, and enzymes (PerkinElmer
Discovery Services Customized Screening Program) showed
no significant inhibitory activity by 38 (Tables S9 and S10). In
addition, compound 38 was not a CYP inhibitor with IC50
values greater than 25 μM for all CYP isoforms tested (1A2,
2B6, 2C8, 2C9, 2C19, 2D6, and 3A4) and showed low hERG
patch clamp activity (9% inhibition at 5 μM). In vitro ADME
profiling showed compound 38 had a high permeability (11 ×
10−6 cm/s) across Caco-2 cell monolayers, suggesting the
potential for high oral absorption. Compound 38 exhibited
moderate metabolic stability in human liver microsomes
(intrinsic clearance: 0.8 L/h/kg) with a free fraction of 11%
in the human plasma protein binding study. Given the good rat
and monkey PK profiles (Table 5), compound 38 was further
evaluated in dogs. When dosed intravenously (1 mg/kg) in
beagle dogs, 38 exhibited low systemic clearance (10% of
hepatic blood flow) with moderate volume distribution (3.49
L/kg), resulting in a long half-life of 15.7 h. Following oral
dosing (2 mg/kg), 38 exhibited high exposure (22100 μM*h)
with almost complete oral bioavailability (98%). The
preclinical PK profiles in rats, monkeys and dogs suggested a
potential of once a day dosing for this compound.
Based on our SAR data and the X-ray cocrystal structure of
PD173074 with FGFR1 (PDB code: 2FGI), we proposed the
binding mode of compound 38 with FGFR1 as shown in
Figure 4. Compound 38 binds in the ATP-binding pocket of
FGFR1 with the pyrrolopyridine fragment making hydrogen
bonds with the NH and CO of Ala564 in the hinge region of
FGFR1 at the distance of approximately 1.8 and 2.0 Å,
respectively. The tricyclic urea scaffold exhibits favorable van
der Waals interactions with the side chains of Val492, Leu484,
and Leu630 with the approximate distance of 3.9, 4.1, and 3.5
Å, respectively, forming sandwich-type hydrophobic interactions. The 2,6-difluoro-3,5-dimethoxylphenyl ring is perpendicular to the tricyclic scaffold and fills a complementary
hydrophobic pocket in the region near the gatekeeper Val561
which is located at approximately 3.6 Å to the phenyl ring. One
methoxy group is at 1.9 Å to the NH of Asp641 and could
potentially form a hydrogen bond. The ethyl group in 38
occupies the same hydrophobic pocket as the ATP ribose. The
morpholine group extends toward the solvent exposed region.
To test the effect of compound 38 on FGFR signaling in
vivo, compound 38 was dosed orally in KatoIII tumor-bearing
mice and it showed dose-proportional plasma exposures after a
single oral dose of 0.1, 1, and 10 mg/kg (Figure S2).31 Based
on these results, a pharmacodynamics (PD) study was
designed to analyze the effect of FGFR inhibition in KatoIII
tumor-bearing mice by compound 38 at dose levels of 0.01−
1.0 mg/kg. The total plasma-exposure and pharmacodynamics
curve indicated an in vivo IC50 of 22 nM for target inhibition
(Figure 5a). The antitumor effect of compound 38 was also
investigated in a mouse xenograft model employing KatoIII
tumor cells. Once-daily oral dosing (0.03−1 mg/kg) achieved
significant and dose-dependent tumor growth inhibition.
Maximum activity was observed with doses equal to or greater
than 0.3 mg/kg (Figure 5b). Throughout the studies, all the
doses were well tolerated with no significant body weight loss
observed. Compound 38 also showed significant in vivo
antitumor activity in several other xenograft mice models with
FGFR genetic alterations, such as KG1 (FGFR1) and RT112
(FGFR3).31 Given the excellent in vitro and in vivo profiles,
compound 38 was progressed into preclinical safety evaluation.
The favorable safety profile in rats and monkeys observed for
Figure 4. Proposed binding mode of compound 38 in the ATP
pocket of FGFR1.
Journal of Medicinal Chemistry pubs.acs.org/jmc Drug Annotation
https://doi.org/10.1021/acs.jmedchem.1c00713
J. Med. Chem. XXXX, XXX, XXX−XXX
I
compound 38 justified its advancement into human clinical
trials.
In the phase I dose escalation study, pemigatinib
(compound 38, INCB054828) exhibited dose-dependent PK
over the dose range of 1−20 mg with the terminal half-life of
15 h supporting once-daily dosing. Sustained inhibition of
FGFR signaling was observed at the RP2D of 13.5 mg once
daily.32 In an open-label single-arm phase II study, among 107
patients with locally advanced or metastatic cholangiocarcinoma harboring FGFR2 fusions or rearrangements whose disease
had progressed following at least one prior therapy, 38 patients
achieved an objective response after treatment with pemigatinib (13.5 mg orally once daily, 2 weeks on/1 week off). The
majority of patients showed evidence of tumor shrinkage after
the treatment (Figure 6). The most common adverse event
was hyperphosphataemia.33 Based on these data, the U.S. Food
and Drug Administration (FDA) granted accelerated approval
to pemigatinib for the treatment of adults with previously
treated, unresectable locally advanced or metastatic cholangiocarcinoma with a FGFR2 fusion or other rearrangement.
Additional clinical studies of pemigatinib in patients with
cancer harboring genetic FGFR alterations are ongoing.
■ CONCLUSIONS
In summary, we have rationally designed a series of tricyclic
urea compounds as potent and selective FGFR inhibitors based
on an in-house screening hit using structure-guided drug
design. By leveraging interactions with the hydrophobic pocket
near the FGFR gatekeeper region, we were able to achieve
good potency and selectivity in this series. Systematic SAR
studies to address low solubility, poor cyno PK and TDI of
CYP3A4 challenges finally led to the discovery of a potent and
highly selective FGFR1, FGFR2, and FGFR3 inhibitor 38
(INCB054828, pemigatinib) with excellent PK profiles and
balanced properties. PK/PD studies of 38 in preclinical
xenograft tumor models confirmed FGFR target engagement,
and strong antitumor effects were observed across multiple
tumor models harboring genetic alterations of FGFR.
Pemigatinib is being studied in multiple clinical trials for
patients with FGFR alterations and has demonstrated a
favorable safety profile with early signs of clinical activity.
FDA has granted accelerated approval to pemigatinib for the
treatment of adults with previously treated, unresectable locally
advanced or metastatic cholangiocarcinoma with a FGFR2
fusion or other rearrangement. Efforts to further demonstrate
the clinical potential of pemigatinib are ongoing.
■ EXPERIMENTAL SECTION
General Synthetic Procedures. All reactions were carried out
under an atmosphere of dry nitrogen. All starting materials and
solvents were used without further purification as acquired from
commercial sources. Purification by flash chromatography was
performed on RediSep columns using Isco CombiFlash SG100c.
Reverse-phase preparative HPLC purifications were performed on
Waters FractionLynx system using UV-triggered or mass directed
fractionation and compound-specific method optimization.34 All final
compounds for biological testing were purified by reverse phase prepHPLC to >95% purity as determined by analytical LC-MS. NMR
spectra were obtained using either a Varian Mercury-300, Mercury-
400, Inova-500, or Bruker Avance-600 MHz spectrometer.
Synthesis of Compound 38 (INCB054828, Pemigatinib):
Step 1. 4-(Ethylamino)-1H-pyrrolo[2,3-b]pyridine-5-carbaldehyde (40). A mixture of 4- chloro-1H-pyrrolo[2,3-b]pyridine-5-
carbaldehyde (3.0 g, 17 mmol) and ethylamine (10 M in water, 8.3
mL, 83 mmol) in 2-methoxyethanol (20 mL, 200 mmol) was heated
to 130 °C and stirred overnight. The mixture was cooled to room
temperature and then concentrated under reduced pressure. The
residue was treated with 1 N HCl (30 mL) and stirred at room
temperature for 1 h and then neutralized with saturated NaHCO3
aqueous solution. The precipitate was collected via filtration and then
washed with water and dried to provide the desired product (2.9 g,
92%) which was used in the next step without further purification.
LC-MS calculated for C10H12N3O [M + H]+ m/z: 190.1; found:
190.1. 1
H NMR (500 MHz, DMSO-d6) δ 11.80 (s, 1H), 9.75 (s, 1H),
9.29 (s, 1H), 8.19 (s, 1H), 7.19 (d, J = 3.6 Hz, 1H), 6.71 (d, J = 3.6
Hz, 1H), 3.77−3.65 (m, 2H), 1.30 (t, J = 7.1 Hz, 3H).
Step 2. 5-{[(2,6-Difluoro-3,5-dimethoxyphenyl)amino]-
methyl}-N-ethyl-1H-pyrrolo[2, 3-b]pyridin-4-amine (42). A
mixture of 4-(ethylamino)-1H-pyrrolo[2,3-b]pyridine-5-carbaldehyde
(7.0 g, 37 mmol), 2,6-difluoro-3,5-dimethoxyaniline (9.1 g, 48 mmol),
and [(1S)-7,7-dimethyl-2-oxobicyclo[2.2.1]hept-1-yl]methanesulfonic
acid (2 g, 7 mmol) in xylenes (250 mL) was heated to reflux with
azeotropic removal of water using a Dean-Stark apparatus for 2 days,
at which time LC-MS analysis showed the reaction was complete.
Then, the mixture was cooled to room temperature and the solvent
was removed under reduced pressure. The residue was dissolved in
tetrahydrofuran (500 mL) and then 2.0 M lithium tetrahydroalumiFigure 5. Data were initially published in ref 31. (a) PK/PD analyses
for compound 38. KatoIII human gastric xenografts were established
in female severe combined immunodeficient (SCID) mice in 5
independent experiments. Once tumors were well established, mice (n
= 3−5/group) were treated with a single oral dose of compound 38 at
dose levels of 0.01, 0.03, 0.1, 0.3, and 1 mg/kg. Four hours after the
single dose, mice were sacrificed and the tumors were collected to
analyze levels of pFGFR2 relative to vehicle treated controls. In
addition, plasma samples were collected for PK analysis. Data were
process from all studies. (b) Efficacy of compound 38 in female SCID
mice bearing KatoIII tumors. Compound 38 was administered orally
once daily for 10 days. The mean tumor size is plotted for each group
of 8 mice. ***P < 0.001 vs vehicle. Adapted from Liu, P. C., et al.
INCB054828 (pemigatinib), a potent and selective inhibitor of
fibroblast growth factor receptors 1, 2 and 3, displays activity against
genetically defined tumor models. PLoS One, 2020, 15(4), e0231877,
10.1371/journal.pone.0231877, under Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/).31
Journal of Medicinal Chemistry pubs.acs.org/jmc Drug Annotation
https://doi.org/10.1021/acs.jmedchem.1c00713
J. Med. Chem. XXXX, XXX, XXX−XXX
J
nate in THF (37 mL, 74 mmol) was added slowly and the resulting
mixture was stirred at 50 °C for 3 h and then cooled to room
temperature. The reaction was quenched by addition of water
followed by 15% aqueous NaOH. The mixture was filtered and
washed with THF. The filtrate was concentrated and the residue was
washed with CH2Cl2 and then filtered to provide the pure product
(11 g, 82%). LC-MS calculated for C18H21F2N4O2 [M + H]+ m/z:
363.2; found: 363.1. 1
H NMR (500 MHz, DMSO-d6) δ 11.03 (s,
1H), 7.62 (s, 1H), 7.09−6.99 (m, 1H), 6.53−6.45 (m, 1H), 6.26 (t, J
= 7.8 Hz, 1H), 6.02−5.91 (m, 1H), 5.41−5.31 (m, 1H), 4.31 (d, J =
6.9 Hz, 2H), 3.74 (s, 6H), 3.66−3.55 (m, 2H), 1.27−1.17 (m, 3H).
Step 3: 3-(2,6-Difluoro-3,5-dimethoxyphenyl)-1-ethyl-
1,3,4,7-tetrahydro-2H-pyrrolo[3′,2’:5,6] pyrido[4,3-d]-
pyrimidin-2-one (43). A solution of triphosgene (5.5 g, 18 mmol)
in tetrahydrofuran (30 mL) was added slowly to a mixture of 5-
{[(2,6- difluoro- 3,5-dimethoxyphenyl)amino]methyl}-N-ethyl-1Hpyrrolo[2,3-b]pyridin-4-amine (5.6 g, 15 mmol) in tetrahydrofuran
(100 mL) at 0 °C, and then the mixture was stirred at room
temperature for 6 h. The mixture was cooled to 0 °C, and then 1.0 M
sodium hydroxide in water (100 mL, 100 mmol) was added slowly.
The reaction mixture was stirred at room temperature overnight. The
formed precipitate was collected via filtration and then washed with
water and dried to provide the first batch of the pure desired product.
The organic layer in the filtrate was separated, and the aqueous layer
was extracted with methylene chloride. The combined organic layer
was concentrated and the residue was triturated with methylene
chloride and then filtered and dried to provide another batch of the
product (total 5.5 g, 92%). LC-MS calculated for C19H19F2N4O3 [M +
H]+ m/z: 389.1; found: 389.1. 1
H NMR (500 MHz, DMSO-d6) δ
11.86 (s, 1H), 7.99 (s, 1H), 7.52−7.46 (m, 1H), 7.04 (t, J = 8.2 Hz,
1H), 6.67−6.62 (m, 1H), 4.76 (s, 2H), 4.18 (q, J = 6.9 Hz, 2H), 3.89
(s, 6H), 1.34 (t, J = 6.9 Hz, 3H).
Step 4: 3-(2,6-Difluoro-3,5-dimethoxyphenyl)-1-ethyl-7-
(phenylsulfonyl)-1,3,4,7-tetrahydro-2H-pyrrolo[3′,2’:5,6]-
pyrido[4,3-d]pyrimidin-2-one (44). To a solution of 3-(2,6-
difluoro-3,5-dimethoxyphenyl)-1-ethyl-1,3,4,7-tetrahydro-2H-pyrrolo-
[3′,2’:5,6]pyrido[4,3-d]pyrimidin-2-one (900 mg, 2.32 mmol) in
N,N-dimethylformamide (20 mL) cooled to 0 °C was added sodium
hydride (185 mg, 4.63 mmol, 60 wt % in mineral oil). The resulting
mixture was stirred at 0 °C for 30 min, and then benzenesulfonyl
chloride (0.444 mL, 3.48 mmol) was added. The reaction mixture was
stirred at 0 °C for 1.5 h, at which time LC-MS analysis showed
completion of the reaction to the desired product. The reaction
mixture was quenched with saturated NH4Cl solution and diluted
with water. The white precipitate was collected via filtration and then
washed with water and hexanes and dried to afford the desired
product (1.2 g, 98%) as a white solid which was used in the next step
without further purification. LC-MS calculated for C25H23F2N4O5S
[M + H]+ m/z: 529.1; found: 529.1. 1
H NMR (600 MHz, DMSO-d6)
δ 8.14−8.10 (m, 2H), 8.09 (s, 1H), 7.94 (d, J = 4.3 Hz, 1H), 7.75−
7.71 (m, 1H), 7.65−7.61 (m, 2H), 7.04 (t, J = 8.1 Hz, 1H), 6.97 (d, J
= 4.3 Hz, 1H), 4.76 (s, 2H), 4.10 (q, J = 7.0 Hz, 2H), 3.88 (s, 6H),
1.28 (t, J = 7.0 Hz, 3H).
Step 5: 3-(2,6-Difluoro-3,5-dimethoxyphenyl)-1-ethyl-2-
oxo-7-(phenylsulfonyl)-2,3,4,7-tetrahydro-1H-pyrrolo-
[3′,2’:5,6]pyrido[4,3-d]pyrimidine-8-carbaldehyde (45). To a
solution of 3-(2,6-difluoro-3,5-dimethoxyphenyl)-1-ethyl-7-(phenylsulfonyl)-1,3,4,7-tetrahydro-2H-pyrrolo[3′,2’:5,6]pyrido[4,3-d]-
pyrimidin-2-one (1.75 g, 3.31 mmol) in tetrahydrofuran (80 mL) at
−78 °C was added freshly prepared lithium diisopropylamide (1 M in
tetrahydrofuran (THF), 3.48 mL, 3.48 mmol). The resulting mixture
was stirred at −78 °C for 30 min, and then N,N-dimethylformamide
(1.4 mL, 18 mmol) was added slowly. The reaction mixture was
stirred at −78 °C for 30 min then quenched with water, warmed to
room temperature, and extracted with EtOAc. The organic extracts
were combined and washed with water and brine. The organic layer
was dried over Na2SO4 and concentrated. The residue was purified by
flash chromatography (0 to 20% EtOAc in CH2Cl2) to give the
desired product as a white solid (1.68 g, 91%). LC-MS calculated for
C26H23F2N4O6S (M+H)+ m/z: 557.1; found: 556.9. 1
H NMR (600
MHz, DMSO-d6) δ 10.45 (s, 1H), 8.26 (s, 1H), 8.25−8.22 (m, 2H),
7.78−7.74 (m, 1H), 7.68−7.63 (m, 2H), 7.62 (s, 1H), 7.05 (t, J = 8.1
Hz, 1H), 4.79 (s, 2H), 4.14 (q, J = 7.0 Hz, 2H), 3.89 (s, 6H), 1.28 (t,
J = 7.0 Hz, 3H).
Step 6: 3-(2,6-Difluoro-3,5-dimethoxyphenyl)-1-ethyl-8-
(morpholin-4-ylmethyl)-7-(phenylsulfonyl)-1,3,4,7-tetrahydro-2H-pyrrolo[3′,2’:5,6]pyrido[4,3-d]pyrimidin-2-one (46). To
a solution 3-(2,6-difluoro-3,5-dimethoxyphenyl)-1-ethyl-2-oxo-7-
(phenylsulfonyl)-2,3,4,7-tetrahydro-1H-pyrrolo[3′,2’:5,6]pyrido[4,3-
d]pyrimidine-8-carbaldehyde (1.73 g, 3.11 mmol) in dichloromethane
(50 mL) was added morpholine (0.95 mL, 11 mmol), followed by
acetic acid (2 mL, 30 mmol). The resulting yellow solution was stirred
at room temperature overnight, and then sodium triacetoxyborohydride (2.3 g, 11 mmol) was added. The mixture was stirred at room
temperature for 3 h, at which time LC-MS showed the reaction went
to completion to the desired product. The reaction was quenched
with saturated NaHCO3 and then extracted with ethyl acetate
(EtOAc). The organic extracts were combined and washed with water
and brine. The organic layer was dried over Na2SO4 and
concentrated. The residue was purified by flash chromatography (0
to 40% EtOAc in CH2Cl2) to give the desired product as a yellow
solid (1.85 g, 95%). LC-MS calculated for C30H32F2N5O6S (M + H)+
Figure 6. Best percentage change from baseline in target lesion size for individual patients with FGFR2 fusions or rearrangements treated with
pemigatinib. (Colored bars indicate confirmed responses accessed by RECIST 1.1. *Patient had a decrease in target lesion size but was not
evaluable for response using RECIST.) Reprinted from Lancet Oncol., 2020, 21, Abou-Alfa, G. K., et al. Pemigatinib for previously treated, locally
advanced or metastatic cholangiocarcinoma: a multicenter, open-label, phase 2 study, 671−684, Copyright (2020), with permission from Elsevier.33
Journal of Medicinal Chemistry pubs.acs.org/jmc Drug Annotation
https://doi.org/10.1021/acs.jmedchem.1c00713
J. Med. Chem. XXXX, XXX, XXX−XXX
K
m/z: 628.2; found: 628.0. 1
H NMR (500 MHz, DMSO-d6) δ 8.45−
8.37 (m, 2H), 8.06 (s, 1H), 7.75−7.69 (m, 1H), 7.68−7.60 (m, 2H),
7.08−7.01 (m, 1H), 6.89 (s, 1H), 4.75 (s, 2H), 4.10 (q, J = 6.9 Hz,
2H), 3.93 (s, 2H), 3.89 (s, 6H), 3.65−3.54 (m, 4H), 2.52−2.49 (m,
4H, overlapped with DMSO), 1.32−1.26 (m, 3H).
Step 7: 3-(2,6-Difluoro-3,5-dimethoxyphenyl)-1-ethyl-8-
(morpholin-4-ylmethyl)-1,3,4,7-tetrahydro-2H-pyrrolo-
[3′,2’:5,6]pyrido[4,3-d]pyrimidin-2-one (38). To a solution of 3-
(2,6-difluoro-3,5-dimethoxyphenyl)-1-ethyl-8-(morpholin-4-ylmethyl)-7-(phenylsulfonyl)-1,3,4,7-tetrahydro-2H-pyrrolo[3′,2’:5,6]-
pyrido[4,3-d]pyrimidin-2-one (1.5 g, 2.4 mmol) in tetrahydrofuran
(40 mL) was added tetra-n-butylammonium fluoride (1 M in THF,
7.2 mL, 7.2 mmol). The resulting solution was stirred at 50 °C for 1.5
h and then cooled to room temperature and quenched with water.
The mixture was extracted with dichloromethane and the organic
extracts were combined and washed with water and brine. The
organic layer was dried over Na2SO4 and concentrated. The residue
was purified by flash chromatography (0 to 10% MeOH in CH2Cl2)
to give the desired product as a white solid, which was further purified
by prep HPLC (pH = 2, acetonitrile/H2O + TFA) to give the desired
product as the TFA salt. LC-MS calculated for C24H28F2N5O4 (M +
H)+ m/z: 488.2; found: 488.0. 1
H NMR (500 MHz, DMSO-d6) δ
12.09 (s, 1H), 8.06 (s, 1H), 7.05 (t, J = 8.1 Hz, 1H), 6.87 (s, 1H),
4.78 (s, 2H), 4.50 (s, 2H), 4.17 (q, J = 6.8 Hz, 2H), 3.97 (br, 2H),
3.89 (s, 6H), 3.65 (br, 2H), 3.37 (br, 2H), 3.15 (br, 2H), 1.37 (t, J =
6.8 Hz, 3H).
■ ASSOCIATED CONTENT
*sı Supporting Information
The Supporting Information is available free of charge at
https://pubs.acs.org/doi/10.1021/acs.jmedchem.1c00713.
Experimental details for in vitro and in vivo studies;
synthetic procedures and characterization for compounds 1−37; in vitro potency, ADME properties and
PK profiles for compounds 2-38; Detailed characterization for compound 38 (PDF)
Molecular formula strings and in vitro biological data
(CSV)
■ AUTHOR INFORMATION
Corresponding Author
Liangxing Wu − Incyte Research Institute, Incyte Corporation,
Wilmington, Delaware 19803, United States; orcid.org/
0000-0001-8524-0799; Email: [email protected]
Authors
Colin Zhang − Incyte Research Institute, Incyte Corporation,
Wilmington, Delaware 19803, United States
Chunhong He − Incyte Research Institute, Incyte Corporation,
Wilmington, Delaware 19803, United States
Dingquan Qian − Incyte Research Institute, Incyte
Corporation, Wilmington, Delaware 19803, United States
Liang Lu − Incyte Research Institute, Incyte Corporation,
Wilmington, Delaware 19803, United States
Yaping Sun − Incyte Research Institute, Incyte Corporation,
Wilmington, Delaware 19803, United States
Meizhong Xu − Incyte Research Institute, Incyte Corporation,
Wilmington, Delaware 19803, United States
Jincong Zhuo − Incyte Research Institute, Incyte Corporation,
Wilmington, Delaware 19803, United States
Phillip C. C. Liu − Incyte Research Institute, Incyte
Corporation, Wilmington, Delaware 19803, United States
Ronald Klabe − Incyte Research Institute, Incyte Corporation,
Wilmington, Delaware 19803, United States
Richard Wynn − Incyte Research Institute, Incyte Corporation,
Wilmington, Delaware 19803, United States
Maryanne Covington − Incyte Research Institute, Incyte
Corporation, Wilmington, Delaware 19803, United States
Karen Gallagher − Incyte Research Institute, Incyte
Corporation, Wilmington, Delaware 19803, United States
Lynn Leffet − Incyte Research Institute, Incyte Corporation,
Wilmington, Delaware 19803, United States
Kevin Bowman − Incyte Research Institute, Incyte
Corporation, Wilmington, Delaware 19803, United States
Sharon Diamond − Incyte Research Institute, Incyte
Corporation, Wilmington, Delaware 19803, United States
Holly Koblish − Incyte Research Institute, Incyte Corporation,
Wilmington, Delaware 19803, United States
Yue Zhang − Incyte Research Institute, Incyte Corporation,
Wilmington, Delaware 19803, United States
Maxim Soloviev − Incyte Research Institute, Incyte
Corporation, Wilmington, Delaware 19803, United States
Gregory Hollis − Incyte Research Institute, Incyte
Corporation, Wilmington, Delaware 19803, United States
Timothy C. Burn − Incyte Research Institute, Incyte
Corporation, Wilmington, Delaware 19803, United States
Peggy Scherle − Incyte Research Institute, Incyte Corporation,
Wilmington, Delaware 19803, United States
Swamy Yeleswaram − Incyte Research Institute, Incyte
Corporation, Wilmington, Delaware 19803, United States
Reid Huber − Incyte Research Institute, Incyte Corporation,
Wilmington, Delaware 19803, United States
Wenqing Yao − Incyte Research Institute, Incyte Corporation,
Wilmington, Delaware 19803, United States
Complete contact information is available at:
https://pubs.acs.org/10.1021/acs.jmedchem.1c00713
Author Contributions
The manuscript was prepared by L.W. with contributions from
all authors.
Notes
The authors declare no competing financial interest.
■ ACKNOWLEDGMENTS
We would like to thank all colleagues at Incyte who were
involved in the pemigatinib project. We thank Karl Blom,
Laurine Galya, and Mei Li for their analytical assistance; Ravi
Jalluri for the molecular modeling; Alex Margulis, Darlise
DiMatteo, Yanlong Li, Xin He, Hong Chang, Kamna Katiyar,
Krista A. Burke, and Ruth Yong-Sciame for their expert
technical assistance; Joe Zhou and his team for the scale up of
pemigatinib; Sarah Winterton, Joshua Hummel, Matthew
McCammant, and Rory McAtee for proofreading the manuscript.
■ ABBREVIATIONS USED
ADME, absorption, distribution, metabolism, and excretion;
AKT, protein kinase B; AUC, area under the curve; AURKB,
Aurora B kinase; cyno, cynomolgus monkey; Cmax, maximum
concentration; FGFR, fibroblast growth factor receptor; FDA,
U.S. Food and Drug Administration; FaSSIF, fasted state
simulated intestinal fluid; hERG, human ether-go-go related
gene; HBF, hepatic blood flow; H-Cl, intrinsic clearance in
human liver microsomes; IV, intravenous; JAK2, Janus kinase
2; c-KIT, proto-oncogene c-KIT receptor tyrosine kinase;
MAPK, mitogen activated protein kinase; PK, pharmacokiJournal of Medicinal Chemistry pubs.acs.org/jmc Drug Annotation
https://doi.org/10.1021/acs.jmedchem.1c00713
J. Med. Chem. XXXX, XXX, XXX−XXX
L
netics; PLCγ, phospholipase C gamma; PDGFRβ, platelet
derived growth factor receptor beta; PI3K, phosphoinositide 3-
kinase; PO, oral administration; PD, pharmacodynamics; RAS,
rat sarcoma virus oncogene; RP2D, recommended phase II
dose; RECIST, response evaluation criteria in solid tumors;
SAR, structure−activity relationship; STAT, signal transducer
and activator of transcription; SGF, simulated gastric fluid;
TRKA, Tropomyosin receptor kinase A; TDI, time dependent
inhibition; t1/2, half-life; VEGFR2, vascular endothelial
growth factor receptor 2; Vdss, volume of distribution at steady
state; WB, whole blood
■ REFERENCES
(1) Eswarakumar, V. P.; Lax, I.; Schlessinger, J. Cellular signaling by
fibroblast growth factor receptors. Cytokine Growth Factor Rev. 2005,
16, 139−149.
(2) Powers, C. J.; McLeskey, S. W.; Wellstein, A. Fibroblast growth
factors, their receptors and signaling. Endocr.-Relat. Cancer 2000, 7,
165−197.
(3) Grose, R.; Dickson, C. Fibroblast growth factor signaling in
tumorigenesis. Cytokine Growth Factor Rev. 2005, 16, 179−186.
(4) Knights, V.; Cook, S. J. De-regulated FGF receptors as
therapeutic targets in cancer. Pharmacol. Ther. 2010, 125, 105−117.
(5) Beenken, A.; Mohammadi, M. The FGF family: biology,
pathophysiology and therapy. Nat. Rev. Drug Discovery 2009, 8, 235−
253.
(6) Turner, N.; Grose, R. Fibroblast growth factor signaling: from
development to cancer. Nat. Rev. Cancer 2010, 10, 116−129.
(7) Macdonald, D.; Reiter, A.; Cross, N. C. The 8p11
myeloproliferative syndrome: a distinct clinical entity caused by
constitutive activation of FGFR1. Acta Haematol. 2002, 107, 101−
107.
(8) Arai, Y.; Totoki, Y.; Hosoda, F.; Shirota, T.; Hama, N.;
Nakamura, H.; Ojima, H.; Furuta, K.; Shimada, K.; Okusaka, T.;
Kosuge, T.; Shibata, T. Fibroblast growth factor receptor 2 tyrosine
kinase fusions define a unique molecular subtype of cholangiocarcinoma. Hepatology 2014, 59, 1427−1434.
(9) van Rhijn, B. W.; van Tilborg, A. A.; Lurkin, I.; Bonaventure, J.;
de Vries, A.; Thiery, J. P.; van der Kwast, T. H.; Zwarthoff, E. C.;
Radvanyi, F. Novel fibroblast growth factor receptor 3 (FGFR3)
mutations in bladder cancer previously identified in non-lethal skeletal
Disorders. Eur. J. Hum. Genet. 2002, 10, 819−824.
(10) Weiss, J.; Sos, M. L.; Seidel, D.; Peifer, M.; Zander, T.;
Heuckmann, J. M.; Ullrich, R. T.; Menon, R.; Maier, S.; Soltermann,
A.; Moch, H.; Wagener, P.; Fischer, F.; Heynck, S.; Koker, M.;
Schottle, J.; Leenders, F.; Gabler, F.; Dabow, I.; Querings, S.;
Heukamp, L. C.; Balke-Want, H.; Ansen, S.; Rauh, D.; Baessmann, I.;
Altmuller, J.; Wainer, Z.; Conron, M.; Wright, G.; Russell, O.;
Solomon, B.; Brambilla, E.; Brambilla, C.; Lorimier, P.; Sollberg, S.;
Brustugun, O. T.; Engel-Riedel, W.; Ludwig, C.; Petersen, I.; Sanger,
J.; Clement, J.; Groen, H.; Timens, W.; Sietsma, H.; Thunnissen, E.;
Smit, E.; Heideman, D.; Cappuzzo, F.; Ligorio, C.; Damiani, S.;
Hallek, M.; Beroukhim, R.; Pao, W.; Klebl, B.; Baumann, M.;
Buettner, R.; Ernestus, K.; Stoelben, E.; Wolf, J.; Nurnberg, P.; Perner,
S.; Thomas, R. K. Frequent and focal FGFR1 amplification associates
with therapeutically tractable FGFR1 dependency in squamous cell
lung cancer. Sci. Transl. Med. 2010, 2, 62ra93.
(11) Kunii, K.; Davis, L.; Gorenstein, J.; Hatch, H.; Yashiro, M.; Di
Bacco, A.; Elbi, C.; Lutterbach, B. FGFR2-amplified gastric cancer cell
lines require FGFR2 and Erbb3 signaling for growth and survival.
Cancer Res. 2008, 68, 2340−2348.
(12) Dieci, M. V.; Arnedoes, M.; Andre, F.; Soria, J. C. Fibroblast
growth factor receptor inhibitors as a cancer treatment: from a
biologic rational to medical perspectives. Cancer Discovery 2013, 3,
264−279.
(13) Andre, F.; Bachelot, T. D.; Campone, M.; Dalenc, F.; PerezGarcia, J. M.; Hurvitz, S. A.; Turner, N. C.; Rugo, H. S.; Shi, M. M.;
Zhang, Y.; Kay, A. C. M.; Yovine, A. J.; Baselga, J. A multicenter,
open-label phase II trial of dovitinib, an FGFR1 inhibitor, in FGFR1
amplified and non-amplified metastatic breast cancer. J. Clin. Oncol.
2011, 29, 508−508.
(14) Jonker, D. J.; Rosen, L. S.; Sawyer, M. B.; de Braud, F.; Wilding,
G.; Sweeney, C. J.; Jayson, G. C.; McArthur, G. A.; Rustin, G.; Goss,
G.; Kantor, J.; Velasquez, L.; Syed, S.; Mokliatchouk, O.; Feltquate, D.
M.; Kollia, G.; Nuyten, D. S. A.; Galbraith, S. A phase I study to
determine the safety, pharmacokinetics and pharmacodynamics of a
dual VEGFR and FGFR inhibitor, brivanib, in patients with advanced
or metastatic solid tumors. Ann. Oncol. 2011, 22, 1413−1419.
(15) Mohammadi, M.; McMahon, G.; Sun, L.; Tang, C.; Hirth, P.;
Yeh, B. K.; Hubbard, S. R.; Schlessinger, J. Structures of the tyrosine
kinase domain of fibroblast growth factor receptor in complex with
inhibitors. Science 1997, 276, 955−960.
(16) Mohammadi, M.; Froum, S.; Hamby, J. M.; Schroeder, M. C.;
Panek, R. L.; Lu, G. H.; Eliseenkova, A. V.; Green, D.; Schlessinger, J.;
Hubbard, S. R. Crystal structure of an angiogenesis inhibitor bound to
the FGF receptor tyrosine kinase domain. EMBO J. 1998, 17, 5896−
5904.
(17) (a) Facchinetti, F.; Hollebecque, A.; Bahleda, R.; Loriot, Y.;
Olaussen, K. A.; Massard, C.; Friboulet, L. Facts and new hopes on
selective FGFR inhibitors in solid tumors. Clin. Cancer Res. 2020, 26,
764−774. (b) Weaver, A.; Bossaer, J. B. Fibroblast growth factor
receptor (FGFR) inhibitors: A review of a novel therapeutic class. J.
Oncol. Pharm. Pract. 2021, 27 (3), 702−710. (c) Krook, M. A.;
Reeser, J. W.; Ernst, G.; Barker, H.; Wilberding, M.; Li, G.; Chen, H.
Z.; Roychowdhury, S. Fibroblast growth factor receptors in cancer:
genetic alterations, diagnostics, therapeutic targets and mechanisms of
resistance. Br. J. Cancer 2021, 124 (5), 880−892.
(18) Lovering, F.; Bikker, J.; Humblet, C. Escape from flatland:
increasing saturation as an approach to improving clinical success. J.
Med. Chem. 2009, 52, 6752−6756.
(19) Gunaydin, H.; Altman, M. D.; Ellis, J. M.; Fuller, P.; Johnson, S.
A.; Lahue, B.; Lapointe, B. Strategy for extending half-life in drug
design and its significance. ACS Med. Chem. Lett. 2018, 9, 528−533.
(20) Smith, D. A.; Beaumont, K.; Maurer, T. S.; Di, L. Relevance of
half-life in drug design. J. Med. Chem. 2018, 61, 4273−4282.
(21) Smith, D. A.; Beaumont, K.; Maurer, T. S.; Di, Li Volume of
distribution in drug design. J. Med. Chem. 2015, 58, 5691−5698.
(22) van de Waterbeemd, H.; Smith, D. A.; Beaumont, K.; Walker,
D. K. Property-based design: optimization of drug absorption and
pharmacokinetics. J. Med. Chem. 2001, 44, 1313−1333.
(23) Veber, D. F.; Johnson, S. R.; Cheng, H. Y.; Smith, B. R.; Ward,
K. W.; Kopple, K. D. Molecular properties that influence the oral
bioavailability of drug candidates. J. Med. Chem. 2002, 45, 2615−
2623.
(24) Zhang, L.; Zhu, H.; Mathiowetz, A.; Gao, H. Deep
understanding of structure-solubility relationships for a diverse set
of organic compounds using matched molecular pairs. Bioorg. Med.
Chem. 2011, 19, 5763−5770.
(25) Walker, M. A. Novel tactics for designing water-soluble
molecules in drug discovery. Expert Opin. Drug Discovery 2014, 9,
1421−1433.
(26) Manallack, D. T.; Prankerd, R. J.; Yuriev, E.; Oprea, T. I.;
Chalmers, D. K. The significance of acid/base properties in drug
discovery. Chem. Soc. Rev. 2013, 42, 485−496.
(27) Ishikawa, M.; Hashimoto, Y. Improvement in aqueous
solubility in small molecule drug discovery programs by disruption
of molecular planarity and symmetry. J. Med. Chem. 2011, 54, 1539−
1554.
(28) St. Jean, D. J. Jr.; Fotsch, C. Mitigating heterocycle metabolism
in drug discovery. J. Med. Chem. 2012, 55, 6002−6020.
(29) Shah, A. K.; Agnihotri, S. A. Recent advances and novel
strategies in pre-clinical formulation development: an overview. J.
Controlled Release 2011, 156, 281−296.
(30) Orr, S. T. M.; Ripp, S. L.; Ballard, T. E.; Henderson, J. L.; Scott,
D. O.; Obach, R. S.; Sun, H.; Kalgutkar, A. S. Mechanism-based
inactiviation (MBI) of cytochrome P450 enzymes: structure-activity
Journal of Medicinal Chemistry pubs.acs.org/jmc Drug Annotation
https://doi.org/10.1021/acs.jmedchem.1c00713
J. Med. Chem. XXXX, XXX, XXX−XXX
M
relationships and discovery strategies to mitigate drug-drug
interaction risks. J. Med. Chem. 2012, 55, 4896−4933.
(31) Liu, P. C.; Koblish, H.; Wu, L.; Bowman, K.; Diamond, S.;
DiMatteo, D.; Zhang, Y.; Hansbury, M.; Rupar, M.; Wen, X.; Collier,
P.; Feldman, P.; Klabe, R.; Burke, K. A.; Soloviev, M.; Gardiner, C.;
He, X.; Volgina, A.; Covington, M.; Ruggeri, B.; Wynn, R.; Burn, T.
C.; Scherle, P.; Yeleswaram, S.; Yao, W.; Huber, R.; Hollis, G.
INCB054828 (pemigatinib), a potent and selective inhibitor of
fibroblast growth factor receptors 1, 2, and 3, displays activity against
genetically defined tumor models. PLoS One 2020, 15 (4),
No. e0231877.
(32) Saleh, M.; Gutierrez, M. E.; Subbiah, V.; Smith, D. C.; Asatiani,
E.; Lihou, C. F.; Zhen, H.; Yeleswaram, S.; Ji, T.; Nemunaitis, J.
Preliminary results from a phase 1/2 study of INCB054828, a highly
selective fibroblast growth factor receptor (FGFR) inhibitor, in
patients with advanced malignancies. Cancer Res. 2017, 77 (13 suppl),
CT111.
(33) Abou-Alfa, G. K.; Sahai, V.; Hollebecque, A.; Vaccaro, G.;
Melisi, D.; Al-Rajabi, R.; Paulson, A. S.; Borad, M.; Gallinson, D.;
Murphy, A. G.; Oh, D.; Dotan, E.; Catenacci, D. V.; Van Cutsem, E.;
Ji, T.; Lihou, C. F.; Zhen, H.; Feliz, L.; Vogel, A. Pemigatinib for
previously treated, locally advanced or metastatic cholangiocarcinoma: a multicenter, open-label, phase 2 study. Lancet Oncol. 2020, 21,
671−684.
(34) Blom, K.; Glass, B.; Sparks, R.; Combs, A. Preparative LC-MS
Purification: Improved Compound Specific Method Optimization. J.
Comb. Chem. 2004, 6, 874−883.
Journal of Medicinal Chemistry pubs.acs.org/jmc Drug Annotation
https://doi.org/10.1021/acs.jmedchem.1c00713
J. Med. Chem. XXXX, XXX, XXX−XXX
N