Inhibitors of epidermal growth factor receptor tyrosine kinase: Optimisation of potency and in vivo pharmacokinetics

Article abstract of DOI:10.1016/j.bmcl.2006.06.054

The structure-activity and structure-property relationships of anilinoquinazoline inhibitors of EGFR were investigated. Strategies to lower volume of distribution and shorten half-life through structure and pK_a modulation are discussed.

Full text of DOI:10.1016/j.bmcl.2006.06.054

Bioorganic & Medicinal Chemistry Letters 16 (2006) 4908–4912
Inhibitors of epidermal growth factor receptor tyrosine kinase:
Optimisation of potency and in vivo pharmacokinetics
Peter Ballard,a Robert H. Bradbury,a Craig S. Harris,b Laurent F. A. Hennequin,b
Mark Hickinson,a Jason G. Kettle,a,* Jane Kendrew,a Teresa Klinowska,a
Donald J. Ogilvie,a Stuart E. Pearson,a Emma J. Williamsa and Ingrid Wilsona
aAstraZeneca, Mereside, Alderley Park, Macclesfield, Cheshire SK10 4TG, UK
bAstraZeneca, Centre de Recherches, Z.I. la Pompelle, BP1050, 51689 Reims Cedex 2, France
Received 30 May 2006; revised 13 June 2006; accepted 14 June 2006
Available online 27 June 2006
Abstract—The structure–activity and structure–property relationships of anilinoquinazoline inhibitors of EGFR were investigated.
Strategies to lower volume of distribution and shorten half-life through structure and pKa modulation are discussed.
ꢀ 2006 Elsevier Ltd. All rights reserved.
In recent years, 4-anilinoquinazolines have emerged as a
versatile template for inhibition of a diverse range of
receptor tyrosine kinases. The most widely studied of
these is the epidermal growth factor receptor (EGFR),1
with the small-molecule inhibitor gefitinib 12 being the
first agent from this class to be approved for the treat-
ment of non-small cell lung cancer refractory to prior
chemotherapeutic intervention. Subsequent research
aimed at further exploration of the SAR of this novel
template has led to compounds that selectively target
an array of different kinases, including for example
SRC,3 aurora,4 the MAP kinase p38 and CDK2.5 Our
own efforts have also identified ZD6474 2,6 a highly po-
tent inhibitor of vascular endothelial growth factor
(VEGF) Receptor and EGFR that is currently in clinical
trials for the treatment of non-small cell lung cancer.7
half-life. Whilst a long terminal half-life is desirable in
terms of increasing drug exposure and ultimately effica-
cy, drug accumulation may in theory be an issue, and
extended washout times required should therapy need
to be discontinued. A detailed DMPK study of gefitinib
indicated half-lives of 7–14 h in rat with corresponding
Vdss = 8–10 l/kg.8 Mean terminal half-life in healthy vol-
unteers is 28 h (range 12–51 h)9 and two separate phase
1 studies in cancer patients have indicated a mean
terminal half-life of 41 h,10 and 48 h, respectively (range
37–65 h).11 In a comparison between ZD6474 and
ZD4190 3, an early prototype VEGFR inhibitor, the
switch to a basic side chain from a neutral one saw the
half-lives increase 12- and 8-fold in rat and dog, respec-
tively.6 In a phase I study in 49 cancer patients, the half-
life of ZD6474 was found to be approximately 120 h.12
Gefitinib 1 and ZD6474 2 share the same anilinoquinaz-
oline core, but differ in both the aniline substitution
pattern, and the nature and position of the side chains.
These basic side chains (gefitinib pKa = 7.2, ZD6474
pKa = 9.3) serve to improve physical properties and con-
fer a favourable pharmacokinetic profile in animals and
human. One consequence of the introduction of a basic
side chain is to increase the volume of distribution at
steady state (Vdss), and thus increase the observed
F
O
HN
N
Cl
N
O
O
N
1 Gefitinib
Br
2 ZD6474,R =
3 ZD4190,R =
HN
N
O
O
F
N
Keywords: EGF; Kinase; Inhibitor; DMPK.
*
N
N
R
Corresponding author. Tel.: +44 1625517920; fax: +44
1625586707; e-mail: jason.kettle@astrazeneca.com
N
N
0960-894X/$ - see front matter ꢀ 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2006.06.054

P. Ballard et al. / Bioorg. Med. Chem. Lett. 16 (2006) 4908–4912
4909
Since treatment outcome for many anti-cancer agents is
often directly linked with pharmacokinetic exposure13
and this exposure can vary substantially between
patients, we sought to re-examine the structure–activity
relationships of gefitinib analogues, in particular with a
view to further optimising potency, and to modulate
half-life through direct targeting of Vdss. Whilst lowering
Vdss should have the desired effect of reducing half-life
(given a constant clearance), at the outset of this work
it was unclear how this strategy would influence physical
properties, and critically, the observed in vivo efficacy,
since effective tissue penetration and duration of effect
will still be required at the site of action. Herein we re-
port our preliminary findings.
resulted in a potent inhibitor of EGFR at both enzyme
and cellular levels.15 Interestingly, improvements were
seen in both solubility and plasma protein binding
(PPB) measurements in both rat and mouse for this
seemingly conservative change. The 6-fluoro analogue
11 was in comparison a weaker inhibitor of EGFR,
although again showed significant increases in solubility
greater than for 10. Protein binding for 11 was also
improved, although in this case slightly inferior to 10
in both species. Table 2 gives selected pharmacokinetic
data for these compounds. In the rat, compared to 1,
both 10 and 11 show improved exposure, driven through
a combination of lower clearance and improved bio-
availability.16 As a precursor to disease model profiling
in mouse xenograft studies, high dose mouse pharmaco-
kinetic studies (po only) were conducted. Again, 10 and
11 showed greater free exposure in the mouse compared
to 1, with 10 in particular showing superior and extended
coverage. By the 24 h time-point however plasma levels
of 10 and gefitinib 1 were similar, and not detected
for 11.
Aniline variants of gefitinib, 10 and 11, were synthesised
by a procedure analogous to that previously reported.14
Scheme 1 outlines the generic synthesis of the remaining
compounds presented in this study. Accordingly,
chloroquinazoline 414 was treated with the requisite flu-
orochloroanilines to give anilinoquinazolines 5, whose
6-phenol could be unmasked by treatment with ammonia
in methanol, to give 6. Alkylation of 6 with protected
side chains was effected under basic conditions, with sul-
fonate esters derived from the corresponding alcohols to
give precursors 7 (n = 0 and 1). Cleavage of the side-
chain carbamates with TFA to yield N–H piperidines
8 was followed by derivatisation, under appropriate
conditions to give target compounds, represented by 9.
We next examined alternative side chains in combina-
tion with these three anilines. Table 3 indicates poten-
cy and physicochemical data for the N-
methylpiperidin-4-ylmethoxy side chain (designated
A, as present in ZD6474) and N-methylpiperidin-4-
yloxy side chain (designated B). Compounds 12 and
13 which both contain the 2-fluoroaniline again show
potent enzyme and cellular inhibition. In particular,
both solubility and rat PPB are improved with side
chain B compared with side chain A. In general, ob-
served solid state melting points for compounds which
contained a secondary alkoxide side chain at C-6,
such as B, were lower than for the corresponding pri-
mary alkoxide side chains such as in A, which we
attribute to less efficient crystal packing. This may
contribute in part to the observed physicochemical dif-
ferences, and it is speculated that a similar phenome-
non may be at work with 2-fluoro analogues 10 and
11 where additional twist in the aniline ring in the sol-
id state leads to improved solubility. For 4-fluoro and
6-fluoro analogues with side chains A and B, activity
appears less consistent, with 15 and 17 (which both
contain N-methylpiperidin-4-yloxy side chain B) show-
ing reduced enzyme and cellular activity.
We first examined alternative fluorine regioisomers of 1,
specifically 2-fluoro analogue 10 and 6-fluoro analogue
11 (Table 1). Introduction of the 2-fluoro substituent
F
O
O
Cl
Cl
N
HN
N
O
O
O
O
a
N
N
4
5
F
F
b
O
O
Cl
Cl
N
HN
N
HN
N
O
HO
O
N
N
c
n
O
d
Compounds 12–17 all share relatively basic side
chains (N-methylpiperidin-4-ylmethoxy side chain A,
pKa = 9.3; N-methylpiperidin-4-yloxy side chain B,
pKa = 8.8) that contribute to the good physicochemical
properties observed. However, the expectation was, as
discussed above, that this may result in an undesirable
pharmacokinetic profile. Table 5 details rat pharmacoki-
netic data for compound 13, which shows a very high
Vdss of 20.8 l/kg consistent with the side-chain basicity,
and high plasma clearance and low total plasma levels,
despite showing good bioavailability.17 The solubility
and free fraction increments observed by switching to
anilines containing a 2-fluoro substituent, as in 10 and
11, combined with the observations around similar
improvements with N-methylpiperidin-4-yloxy side
chain B, led us to target a small array of compounds
6
7 (n = 0,1)
F
F
R
Cl
N
Cl
HN
HN
HN
N
O
O
e
N
N
n
n
O
N
O
8
9
Scheme 1. Generic synthesis scheme for 6-substituted anilinoquinaz-
olines. Reagents and conditions: (a) fluorochloroaniline, IPA, reflux;
(b) aq NH3, MeOH, 50 ꢁC; (c) n = 0: R-OMs, CsF, DMA, 85 ꢁC;
n = 1: R-OTos, K2CO3, DMA, 80 ꢁC; (d) TFA, 25 ꢁC; (e) R = Me: aq
formaldehyde, HCO2H, 90 ꢁC; R = CH2CH2OMe: K2CO3, DMA,
60 ꢁC; R = CH2CONH2: DIPEA, DCM, 25 ꢁC; R = SO2Me: DIPEA,
DCM, 25 ꢁC.

4910
P. Ballard et al. / Bioorg. Med. Chem. Lett. 16 (2006) 4908–4912
Table 1. Potency and physicochemical properties of fluorine regioisomers of gefitinib
F
6
4
O
HN
N
Cl
2
N
O
O
N
Solubilityb (lM)
Rat PPB %
free drugc
Mouse PPB %
free drugc
a
EGFR IC50 (nM)
a
KB cell IC50 (nM)
Compound
1
4-F
2-F
6-F
23d
80d
21
4
50
3.3
7.2
5.3
3.0
7.5
7.3
10
11
18
51
5
4
9
110 37
320
a For determinations where n P 2, standard deviation is given.
b Aqueous solubility measured in pH 7.4 buffer on pure but generally amorphous material.
c Measured at 37 ꢁC.
d See Ref. 14.
Table 2. In vivo pharmacokinetics of fluorine regioisomers of gefitinib in rodents
b
Compound
Rat pharmacokinetic profilea
Mouse free plasma levels (lM)
Bioavailability (%)
Cl (ml/min/kg)
Vdss (l/kg)
AUC (po)c (lM h)
2 h
4 h
6 h
24 h
1
4-F
2-F
6-F
56
73
71
68
29
16
5.8
4.3
3.5
0.32
0.94
1.64
0.26
0.43
0.30
0.13
0.68
0.36
0.15
0.41
0.17
0.01
0.01
0
10
11
a From an oral dose of 10 mg/kg, iv dose of 2 mg/kg in Alderley Park—Wistar rats.
b Following a single oral dose of 50 mg/kg in Alderley Park mice. Free levels calculated on the basis of data in Table 1.
c Total AUC normalised to a dose of 1 mg/kg.
Table 3. Potency and physicochemical data for anilinoquinazolines 12–17
F
6
4
HN
N
Cl
2
R =
O
O
N
N
R
N
A
B
Solubilityb (lM)
Rat PPB %
free drugc
a
EGFR IC50 (nM)
a
KB cell IC50 (nM)
Compound
Aniline
R
12
13
14
15
16
17
2-F
2-F
4-F
4-F
6-F
6-F
A
B
A
B
A
B
2
2
0
2
2
2
2
15
21
6
1
7
15
130
120
100
1885
—
11.2
18.3
2.5
8.2
8.6
—
3
18
10
9
300
58
4
106 15
115 54
a For determinations where n P 2, standard deviation is given.
b Aqueous solubility measured in pH 7.4 buffer on pure but generally amorphous material.
c Measured at 37 ꢁC.
based upon reduction of the pKa in this series. It was
anticipated that the likely reduction in solubility and
free drug levels through lowering of pKa could at least
in part be ameliorated by these beneficial structural
changes. Accordingly, we selected three N-substituents
to span a range of lowered side-chain basicities, an
N-methoxyethoxy (pKa = 8.1, compounds 18 and 21),
and N-2-acetamide (pKa = 6.7, compounds 19 and 22)
and an N–SO2Me (neutral, compounds 20 and 23) side
chain (Table 4).
As before, compounds containing the 2-fluoro aniline
showed greater potency at enzyme and cellular levels
than their 6-fluoro counterparts. Activity was fairly con-
sistent across the various N-substituents for a given ani-
line, although the more basic analogues 18 and 21 did
show reduced activity, particularly in cells. Compounds
19 and 20 are particularly noteworthy for their very po-
tent inhibition of EGFR in these assays. The examina-
tion of the effects on solubility and free fraction for a
particular triad in this set is instructive. Free fraction

P. Ballard et al. / Bioorg. Med. Chem. Lett. 16 (2006) 4908–4912
4911
is seen to progressively decrease moving from basic to
neutral with both anilines (compare the transition 18
through 20, and 21 through 23). A similar effect is seen
in solubility for the 6-fluoro analogues 21–23, where sol-
ubility drops from 210 lM for the more basic 21 to
1 lM for the neutral 23. This effect is not observed with
the 2-fluoroanilino analogues 18–20, where solubility re-
mains modest over the pKa range.
basic close analogue 13. Bioavailability was observed to
be modest, perhaps reflecting solubility limited absorp-
tion. Nevertheless, oral exposure was high, as clearance
is significantly lowered in this set. Taking all this data
together, compound 20 emerged as the best balance of
potency, DMPK and physical properties and was select-
ed for further profiling in a mouse LoVo xenograft
study. One question regarding the targeting of Vdss as
a strategy to reduce half-life is the effect on achieving
sufficient and extended coverage in disease models.
For compound 20, the observed half-life in rat is 1.6 h,
demonstrating the impact of lowering Vdss in this series,
through targeting of lower pKa.
Based on the superior potency seen throughout this
study with 2-fluoroanilino analogues, and the
acceptable physicochemical properties observed with re-
duced side-chain pKas seen for compounds 18–20 in par-
ticular, these compounds were progressed into a rat
pharmacokinetic study (Table 5). All three compounds,
as anticipated, exhibit a lower Vdss relative to the highly
We elected to examine 20 in the disease model with
both once, and twice daily dosing to examine the
Table 4. Potency and physicochemical data for anilinoquinazolines 18–23
F
6
HN
N
Cl
2
O
N
N
R
O
Solubilityb (lM)
Rat PPB %
free drugc
a
EGFr IC50 (nM)
Compound
Aniline
R
KB cell
a
IC50 (nM)
18
19
20
21
22
23
2-F
2-F
2-F
6-F
6-F
6-F
CH2CH2OMe
CH2CONH2
SO2Me
3
2
1
1
2
1
2
0
35 30
3.5
1.7
4
13.4
4.4d
3.3
4
7
3
4
2
CH2CH2OMe
CH2CONH2
SO2Me
114
9
166
10 11
41 13
210
8.9
1
9.9
3.1
1
1.8
a For determinations where n P 2, standard deviation is given.
b Aqueous solubility measured in pH 7.4 buffer on pure but generally amorphous material.
c Measured at 37 ꢁC.
d Measured at 25 ꢁC.
Table 5. Selected rat pharmacokinetic dataa for anilinoquinazolines 13, and 18–20
Compound
R
Side-chain pKa
Bioavailability (%)
Cl (ml/min/kg)
Vdss (l/kg)
AUC
(po)b (lM h)
13
18
19
20
Me
8.8
8.1
6.7
—
74
30
27
45
129
11
20.8
1.3
2.7
1.5
0.08
0.95
0.59
1.18
CH2CH2OMe
CH2CONH2
SO2Me
16
13
a From an oral dose of 5 mg/kg, iv dose of 2 mg/kg in Alderley Park—Wistar rats.
b Total AUC normalised to a dose of 1 mg/kg.
Table 6. Comparison of xenograft data for compounds 1 and 20
c
Compound
Dose (mg/kg)
Inhibition of
tumour volumea (%)
Significance
(P value)
AUC (po)c
Cmax (lM)
(lM h)
1
1
50
100
60
84
53
60
91
0.018
0.002
0.018
78.7
85.9
8.1
6.7
20
20
20b
50
100
224.9
450.5
286.3
30.6
36.8
23.9
0.011
<0.001
2 · 50
a Inhibition of tumour volume relative to control group measured at day 17.
b Doses separated by 12 h.
c Exposure observed in xenograft study.

4912
P. Ballard et al. / Bioorg. Med. Chem. Lett. 16 (2006) 4908–4912
6. Hennequin, L. F.; Stokes, E. S. E.; Thomas, A. P.;
impact of these changes. Female nude mice were
implanted subcutaneously with 1 · 107 LoVo colon
tumour cells. Once tumours were established, animals
were randomised between groups (n = 7 per group)
and dosed po with compound suspended in 1% (v/v)
polysorbate 80. Table 6 indicates the efficacy and
exposure of compound 20 in this study, alongside 1
for comparative purposes. As expected, gefitinib 1
shows good activity and clear dose–response at 50
and 100 mg/kg doses. Compound 20 shows compara-
ble activity at 50 mg/kg, although appears less
efficacious at the higher dose of 100 mg/kg, despite
linear increases in exposure over the dose range.
However, consistent with the shorter half-life for this
´
Johnstone, C.; Ple, P. A.; Ogilvie, D. J.; Dukes, M.;
Wedge, S. R.; Kendrew, J.; Curwen, J. O. J. Med. Chem.
2002, 45, 1300.
7. Heymach, J. V. Br. J. Cancer 2005, 92(Suppl.1), S14.
8. McKillop, D.; Partridge, E. A.; Hutchison, M.; Rhead, S.
A.; Parry, A. C.; Bardsley, J.; Woodman, H. M.; Swais-
land, H. C. Xenobiotica 2004, 34, 901.
9. Swaisland, H.; Laight, A.; Stafford, L.; Jones, H.; Morris,
C.; Dane, A.; Yates, R. Clin. Pharmacokinet. 2001, 40,
297.
10. Tiseo, M.; Loprevite, M.; Ardizzoni, A. Curr. Med.
Chem.—Anti-Cancer Agents 2004, 4, 139.
11. Ranson, M.; Hammond, L. A.; Ferry, D.; Kris, M.; Tullo,
A.; Murray, P. I.; Miller, V.; Averbuch, S.; Ochs, J.;
Morris, C.; Feyereislova, A.; Swaisland, H.; Rowinsky, E.
K. J. Clin. Oncol. 2002, 20, 2240.
compound, dosing twice
a day results in 91%
inhibition of tumour volume relative to control
animals, substantially better than once a day dosing,
including at 100 mg/kg.
12. Miller, K. D.; Trigo, J. M.; Wheeler, C.; Barge, A.;
Rowbottom, J.; Sledge, G.; Baselga, J. Clin. Cancer Res.
2005, 11, 3369.
13. Undevia, S. D.; Gomez-Abuin, G.; Ratain, M. J. Nat. Rev.
Cancer 2005, 5, 447.
14. Barker, A. J.; Gibson, K. H.; Grundy, W.; Godfrey, A. A.;
Barlow, J. J.; Healy, M. P.; Woodburn, J. R.; Ashton, S.
E.; Curry, B. J.; Scarlett, L.; Henthorn, L.; Richards, L.
Bioorg. Med. Chem. Lett. 2001, 11, 1911.
15. Compounds were evaluated in an EGFR kinase assay
measuring inhibition of phosphorylation of a synthetic
peptide substrate at Km ATP concentration. Inhibition of
proliferation of KB cells in response to EGF stimulus was
also assessed. For details of assay conditions see, Henne-
quin, L. F. A.; Kettle, J. G.; Pass, M.; Bradbury, R. H.
PCT Int. Appl. WO2003040109.
In summary, in the compounds tested aniline and side-
chain structural modifications to the EGFR inhibitory
anilinoquinazoline template gave substantial increases
in enzyme and cellular potencies. These changes in com-
bination with side-chain pKa modulation have led to
compounds with a markedly altered pharmacokinetic
profile, which nevertheless show profound inhibition
of tumour growth in a selected disease model.
References and notes
16. The rat pharmacokinetic data reported in this paper are
from a protocol designed to allow high throughput
profiling of compounds and as such are non-optimised.
Literature data for 1 (see Ref. 8) show some variance with
the data in Table 2 due to the more detailed nature, and
differing protocols used in these studies.
1. Hynes, N. E.; Lane, H. A. Nat. Rev. Cancer 2005, 5, 341.
2. Cohen, M. H.; Williams, G. A.; Sridhara, R.; Chen, G.;
McGuinn, W. D., Jr.; Morse, D.; Abraham, S.; Rahman,
A.; Liang, C.; Lostritto, R.; Baird, A.; Pazdur, R. Clin.
Cancer Res. 2004, 10, 1212.
17. While the clearance figure of 129 ml/min/kg is seemingly at
odds with the reported bioavailability of 74%, it should be
noted that this is for clearance of drug from plasma. This
basic compound partitions extensively into blood, such
that the blood clearance is a more realistic 65 ml/min/kg.
Nevertheless, the good bioavailability observed may
reflect saturation of first-pass clearance, or other non-
hepatic routes of elimination.
3. Ple, P. A.; Green, T. P.; Hennequin, L. F.; Curwen, J.;
Fennell, M.; Allen, J.; Lambert-van der Brempt, C.;
Costello, G. J. Med. Chem. 2004, 47, 871.
4. Mortlock, A. A; Keen, N. J.; Jung, F. H.; Brewster, A. G.
PCT Int. Appl. WO 2001021596.
5. Shewchuk, L.; Hassell, A.; Wisely, B.; Rocque, W.;
Holmes, W.; Veal, J.; Kuyper, L. F. J. Med. Chem.
2000, 43, 133.