Optimization of pharmacokinetics through manipulation of physicochemical properties in a series of HCV inhibitors

Article abstract of DOI:10.1021/ml200163b

A novel series of HCV replication inhibitors based on a pyrido[3,2-d]pyrimidine core were optimized for pharmacokinetics (PK) in rats. Several associations between physicochemical properties and PK were identified and exploited to guide the design of comp

Full text of DOI:10.1021/ml200163b

LETTER
pubs.acs.org/acsmedchemlett
Optimization of Pharmacokinetics through Manipulation of
Physicochemical Properties in a Series of HCV Inhibitors
Scott E. Lazerwith,*,† Gina Bahador,§ Eda Canales,† Guofeng Cheng,§ Lee Chong,† Michael O. Clarke,†
Edward Doerffler,† Eugene J. Eisenberg,‡ Jaclyn Hayes,‡ Bing Lu,‡ Qi Liu,† Mike Matles,‡
Michael Mertzman,† Michael L. Mitchell,† Philip Morganelli,† Bernard P. Murray,‡ Margaret Robinson,§
Robert G. Strickley,|| Megan Tessler,|| Neeraj Tirunagari,§ Jianhong Wang,‡ Yujin Wang,‡
Jennifer R. Zhang,† Xubin Zheng,‡ Weidong Zhong,§ and William J. Watkins†
†Medicinal Chemistry, ‡Drug Metabolism, §Biology, and Formulation and Process Development, Gilead Sciences, Inc.,
333 Lakeside Drive, Foster City, California 94404, United States
S Supporting Information
b
ABSTRACT: A novel series of HCV replication inhibitors
based on a pyrido[3,2-d]pyrimidine core were optimized
for pharmacokinetics (PK) in rats. Several associations bet-
ween physicochemical properties and PK were identified and
exploited to guide the design of compounds. In addition, a
simple new metric that may aid in the prediction of bioavail-
ability for compounds with higher polar surface area is described
(3*HBD-cLogP).
KEYWORDS: Pharmacokinetics, HCV inhibitors, pyrido[3,2-d]pyrimidine core, 3*HBD-cLogP
ince the pioneering work by Lipinski and co-workers corre-
lating the physical properties of a compound [molecular
optimizing their physicochemical properties. By both lowering the
PSA and raising cLogP, compounds were discovered with favor-
able potency and rat PK profiles. In addition, we note the difficulty
in predicting the bioavailability of compounds with moderately
high PSA and the significantly negative impact that HBDs appear
to impart within this subgroup.
S
weight, lipophilicity, and the numbers of hydrogen bond donors
and acceptors (HBDs and HBAs)] with good absorption
properties,1 there has been significant effort within the field of
medicinal chemistry to further define and develop methods that allow
for the prediction of pharmacokinetics (PK) (especially oral
bioavailability) from simple calculated properties. For example,
Veber et al. analyzed a large and diverse set of compounds and
proposed two separate rules based on either polar surface area
(PSA) and rotatable bond count or HBDs and HBAs and
rotatable bond count.2 Leeson and Davis compared the change
in physical properties for approved oral drugs before and after
1983 and concluded that the most important oral druglike
physical properties are lipophilicity, HBDs, and percent PSA.3
Johnson et al. proposed a model for optimizing permeability and
clearance based on molecular weight and lipophilicity.4 Other
contributions include a decision tree model based on logD, PSA,
molecular weight, and number of HBDs,5 and a bioavailability
scoring model based on PSA, charge, and the “rule-of-five”.6
Many other more complex prediction tools have been proposed,7
including a “statistical pattern recognition model” based on PSA
and AlogP98.8
Consistent with the literature,2 compounds with PSA > 140
within this series tended to exhibit poor bioavailability, as illu-
strated by Figure 1 (in which the coloring scheme is chosen for
ease of comparison with subsequent figures).10 For example, the
replacement of fluorine by triazole (3 vs 4, Figure 2) results in a
marked reduction in bioavailability.
The relationship between low bioavailability and PSA is
understood to be the result of poor passive permeability of more
polar compounds across intestinal membranes.2,12 Because the
cutoff for good predicted bioavailability is PSA < 140, it was
somewhat surprising that a significant portion of the compounds
in the PSA 140À160 range exhibited moderate to good bioavail-
ability (see below).
A second notable feature of this series was an association
between lipophilicity and mean residence time (MRT). Long
MRTs were rare in compounds with cLogP < 3. Conversely,
compounds of greater lipophilicity (cLogP > 4) were much more
likely to have a long MRT in the rat.13
Chronic hepatitis C virus (HCV) infection represents a major
unmet medical need with an estimated 130À170 million people
affected worldwide.9 To address this, we recently began investi-
gating a series of replication inhibitors based on a weakly basic
2-amino-pyrido[3,2-d]pyrimidine core. Herein, we show how
the PK of compounds in this series was improved by analyzing and
Received: January 14, 2011
Accepted: July 31, 2011
Published: August 08, 2011
r
2011 American Chemical Society
715
dx.doi.org/10.1021/ml200163b ACS Med. Chem. Lett. 2011, 2, 715–719
|

ACS Medicinal Chemistry Letters
LETTER
Figure 1. Relationship between rat bioavailability and PSA.
Figure 3. Relationship between MRT in the rat and cLogP.
Table 1. cLogP and Observed PK Parameters for
Compounds 5, 6, 3, and 7
predicted rat CL
(L/h/kg)
rat CL
rat Vss rat MRT
compd no. cLogP
(L/h/kg) (L/kg)
(h)
5
6
3
7
2.8
3.8
4.7
6.0
<0.34
0.46
0.9
0.2
0.3
0.4
0.9
1
0.9
4.9
6.9
11
<0.34
<0.34
2.3
4.3
Figure 2. Example of increased PSA leading to reduced bioavailability.11
Compounds 5À8 (Figure 4) illustrate this trend. Small
changes (even in different parts of the molecule) that increase
lipophilicity also significantly increase the MRT. Within this set,
cLogP ranges from 2.8 to 6.0, and the MRT increases from less
than 1 h to a very promising 11 h.
In the course of this optimization, it became evident that a
number of compounds with PSA 140À160 still exhibited mod-
erate bioavailability and that PSA alone was a poor predictor of
bioavailability for this subset (Figure 1, red dots).
Because MRT is a function of volume and clearance, the
dependence of both on lipophilicity for this series was examined.
No broad relationship emerged, although the data suggest that
lipophilicity may, in some cases, be influencing volume, at least in
the higher cLogP ranges. For example, in the series of compounds
depicted in Figure 4, the increase in MRT is attributable to higher
volumes (substantially above total body water) for the more
lipophilic compounds 3 and 7. In contrast, the increase in MRT
between compounds 5 and 6 is attributable entirely to reduced
clearance, which was not predicted by hepatic microsomal stability
(Table 1). Thus, in some cases within this data set and consistent
with literature precedent,14 higher lipophilicity is associated with
higher volume (due to increased tissue distribution). At the lower
end of the cLogP range, it appears that further clearance mechan-
isms not predicted by hepatic microsomal stability assays (such as
biliary excretion or renal elimination) are operative.
Through simultaneous manipulation of PSA and cLogP,
analogues were identified with dramatically improved rat PK pro-
files without significant compromise in antiviral potency. Figure 5
illustrates two separate series in which controlled reduction of PSA
and increase in cLogP resulted in two compounds with excellent
overall characteristics. The rat plasma exposure of one such
optimized compound (2) is illustrated in Figure 6.
The presence of HBDs in drug candidates has frequently been
cited as an important factor in bioavailability prediction, with
high HBD count causing a significant impediment to good oral
bioavailability.1À3,5 Indeed, the relative weighting of HBDs and
HBAs in Lipinski's “rule of 5” (maximum of five HBDs and
maximum of 10 HBAs) suggests that even small increases in HBD
count can result in detrimental effects on absorption.1 Consistent
with this paradigm, the bioavailability of compounds in this series
with PSA 140À160 appears to be significantly influenced by
HBD (as opposed to HBA) count (Figures 7 and 8).15 Still, HBD
count alone is insufficient for the prediction of bioavailability.
A comparison of 12 and 13 (Figure 9) suggests that lipophi-
licity might also influence bioavailability for compounds in this
set. The bioavailability of 12, an analogue with high PSA (160)
and HBD count (4), was very low.16 The introduction of two
trifluoromethyl groups in 13, with the attendant increase in cLogP
from 4.3 to 6.5, provided the desired improvement in % F.17
Thus, we became interested in exploring combinations of
descriptors for both lipophilicity and polarity for the prediction
of bioavailability of compounds within this series with PSA just
above the range previously identified as preferred.2
Empirically, we found the weighted combination of HBD
count and cLogP shown in Figure 10 (3*HBD-cLogP) to serve
716
dx.doi.org/10.1021/ml200163b |ACS Med. Chem. Lett. 2011, 2, 715–719

ACS Medicinal Chemistry Letters
LETTER
Figure 4. Examples of the association between MRT in the rat
and cLogP.
Figure 7. Relationship between % F (rat) and HBA count for analogues
with PSA 140À160.
Figure 8. Relationship between % F (rat) and HBD count for analogues
with PSA 140À160.
Figure 5. Two subseries in which PK was optimized through manip-
ulation of PSA and cLogP.
the purpose well: 10 out of 15 compounds with a 3*HBD-
cLogP < 6 showed acceptable bioavailability (>20%), while all
10 compounds with a 3*HBD-cLogP > 6 showed poor
bioavailability.
Applying this metric to all of the compounds in the series,
a similar and dramatic drop in % F is seen just above 3*HBD-
cLogP = 6 (Figure 11). Similar combinations of HBD count and
cLogP as predictors of bioavailability may be of value in other
chemical series.
The significance of the metric 3*HBD-cLogP merits consid-
eration. It is likely that in this “border-region” of high
PSA compounds, low permeability limits bioavailability. Although
Figure 6. IV and oral rat PK for compound 2.
717
dx.doi.org/10.1021/ml200163b |ACS Med. Chem. Lett. 2011, 2, 715–719

ACS Medicinal Chemistry Letters
LETTER
permeability was not routinely measured in this study, the
PAMPA Papp (apparent permeability) of 13 (Figure 9) was
0.09 Â 10À6 cm/s, whereas that of the less bioavailable 12 was
0.004 Â 10À6 cm/s. The 3-fold weighting of HBDs in this metric
emphasizes its influence upon permeability and implies that the
detrimental effect of an extra HBD is compensated for only by
the addition of a large amount of lipophilicity. This study
demonstrates the value of analyzing different components of
PSA and suggests that an alternative calculation of molecular
polarity incorporating the particular significance of HBDs would
provide a better predictor of permeability (and bioavailability).
In summary, analysis of the associations between physicochem-
ical properties and pharmacokinetics in a series of pyrido[3,2-d]-
pyrimidine HCV replication inhibitors resulted in three notable
trends: (i) Compounds with high PSA (PSA >160) were unlikely
to show good bioavailability in the rat, (ii) more lipophilic
compounds generally had longer MRTs, and (iii) compounds
with moderately high PSA (140À160) may show acceptable
bioavailability if they are lipophilic enough. By exploiting these
trends, several compounds (7, 10, and 2) with favorable rat PK
profiles were discovered. Finally, the weighted combination of HBD
count and cLogP (3*HBD-cLogP) was found to be a better
predictor of bioavailability for the moderately high PSA compounds
(140À160) in this series than any single descriptor.
Figure 9. Example of a pair of high PSA analogues in which % F was
improved by an increase in lipophilicity.
’ ASSOCIATED CONTENT
S
Supporting Information. Experimental procedures for
b
assay protocols, in vivo studies, synthesis, and characterization of
compounds 1À13, as well as a complete set of calculated physico-
chemical properties and measured PK parameters for >180 com-
pounds within the series described herein. This material is available
free of charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION
Figure 10. Relationship between % F (rat) and 3*HBD-cLogP for
analogues with PSA 140À160.
Corresponding Author
*Tel: 650-522-1564. E-mail: scott.lazerwith@gilead.com.
’ ACKNOWLEDGMENT
We thank Kathy Brendza for high-resolution mass spectro-
metry and Michael Zhang for help with calculated properties.
’ REFERENCES
(1) Lipinski, C. A.; Lombardo, F.; Dominy, B. W.; Feeney, P. J.
Experimental and computational approaches to estimate solubility and
permeability in drug discovery and development settings. Adv. Drug
Delivery Rev. 1997, 23, 4–25.
(2) 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.
(3) Leeson, P. D.; Davis, A. M. Time-Related Differences in the
Physical Property Profiles of Oral Drugs. J. Med. Chem. 2004, 47,
6338–6348.
(4) Johnson, T. W.; Dress, K. R.; Edwards, M. Using the Golden
Triangle to Optimize Clearance and Oral Absorption. Bioorg. Med.
Chem. Lett. 2009, 19, 5560–5564.
(5) Hou, T.; Wang, J.; Zhang, W.; Xu, X. ADME Evaluation in Drug
Discovery. 7. Prediction of Oral Absorption by Correlation and Classi-
fication. J. Chem. Inf. Model. 2007, 47, 208–218.
Figure 11. Relationship between % F (rat) and 3*HBD-cLogP for all of
the pyridopyrimidines.
(6) Martin, Y. C. A Bioavailability Score. J. Med. Chem. 2005,
48, 3164–3170.
718
dx.doi.org/10.1021/ml200163b |ACS Med. Chem. Lett. 2011, 2, 715–719

ACS Medicinal Chemistry Letters
LETTER
(7) For a review, see Egan, W. J.; Lauri, G. Prediction of intestinal
permeability. Adv. Drug Delivery Rev. 2002, 54, 273–289.
(8) Egan, W. J.; Merz, K. M., Jr.; Baldwin, J. J. Prediction of Drug
absorption Using Multivariate Statistics. J. Med. Chem. 2000, 43, 3867–3877.
(9) Baldo, V.; Baldovin, T.; Trivello, R.; Floreani, A. Epidemiology of
HCV Infection. Curr. Pharm. Des. 2008, 14, 1646–1654.
(10) When bioavailability is normalized to remove the predicted effects
of first pass metabolism [% F/(1 À Cl/QH)], the same trend is seen.
(11) RLMis ratlivermicrosomalstability, quantifiedasinvitrohalf-life.
(12) Clarke, D. E. Rapid Calculation of Polar Molecular Surface Area
and Its Application to the Prediction of Transport Phenomena. 1.
Prediction of Intestinal Absorption. J. Pharm. Sci. 1997, 88, 807–814
and references therein.
(13) This trend holds for compounds in this series with low
predicted clearance based on rat liver microsomes.
(14) Obach, R. S. Prediction of Human Volume of Distribution
Using in vivo, in vitro, and in silico Approaches. In Annual Reports in
Medicinal Chemistry; Macor, J. E., Ed.; Academic Press: London, United
Kingdom, 2007; pp 469À488.
(15) Similar analyses comparing % F with molecular weight and
number of rotatable bonds for this subgroup also failed to produce useful
relationships.
(16) Although microsomal stability differences between 12 and 13
would account for some of the bioavailability difference between these
two compounds, the modest predicted metabolic stability in compound
12 should only reduce the observed % F by about 1/3 based on first-pass
metabolism. Thus, metabolism in the liver is unlikely to be a significant
cause of the large bioavailability difference seen here.
(17) Raising lipophilicity significantly as a strategy for the improve-
ment of bioavailability should be approached with caution, as compounds
with high cLogP also tend towards other undesirable properties, such as
poor solubility and poor selectivity. See Leeson, P. D.; Springthorpe, B.
The influence of drug-like concepts on decision-making in medicinal
chemistry. Nature Rev. Drug Discovery 2007, 6, 881–890.
719
dx.doi.org/10.1021/ml200163b |ACS Med. Chem. Lett. 2011, 2, 715–719