Central Nervous System Multiparameter Optimization Desirability

Subscriber access provided by Lulea University of Technology

Article

Central Nervous System Multi-Parameter Optimization (CNS MPO) Desirability: Application in Drug Discovery Travis T Wager, Xinjun Hou, Patrick R. Verhoest, and Anabella Villalobos ACS Chem. Neurosci., Just Accepted Manuscript • DOI: 10.1021/acschemneuro.6b00029 • Publication Date (Web): 18 Mar 2016 Downloaded from http://pubs.acs.org on March 21, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

ACS Chemical Neuroscience is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 34

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Chemical Neuroscience

Central Nervous System Multi-Parameter Optimization (CNS MPO) Desirability: Application in Drug Discovery Travis T. Wager†, Xinjun Hou, Patrick R. Verhoest and Anabella Villalobos Worldwide Medicinal Chemistry, Pfizer Worldwide Research and Development, 610 Main St Cambridge, Massachusetts 02139

ABSTRACT: Significant progress has been made in prospectively designing molecules using the CNS MPO desirability tool, as evidenced by the analysis reported herein of a second wave of drug candidates that originated after the development and implementation of this tool. This simple-touse design algorithm has expanded design space for CNS candidates, and has further demonstrated the advantages of utilizing a flexible, multi-parameter approach in drug discovery, rather than individual parameters and hard cutoffs of physicochemical properties. The CNS MPO tool has helped increase the percentage of compounds nominated for clinical development that exhibit alignment of ADME attributes, cross the blood-brain barrier, and reside in lower-risk safety space (low ClogP and high TPSA). The use of this tool has played a role in reducing the number of compounds submitted to exploratory toxicity studies and increasing the survival of our drug candidates through regulatory toxicology into First in Human (FIH) studies. Overall, 1 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 34

the CNS MPO algorithm has helped to improve the prioritization of design ideas and the quality of the compounds nominated for clinical development.

KEYWORDS: Attrition, Central nervous system (CNS), CNS candidates, CNS drugs, CNS drug design, CNS MPO, Desirability score, Efficacious drug concentration (Ceff), Harrington optimization, Human liver microsome stability, Hydrogen bond donor, Lipophilicity, MadinDarby canine kidney, Molecular weight, Most basic pKa, Multi-parameter optimization (MPO), Multi-variant optimization, Passive permeability, P-glycoprotein (P-gp), Polarity, Topological polar surface area, Unbound intrinsic clearance. INTRODUCTION In an effort to reduce attrition of our clinical candidates and prospectively increase our odds of success, we began to consider alternative ways to assess the quality of our design ideas. In 2010 we reported our initial analysis of CNS drug property space, which included examination of physicochemical, in vitro ADME (Absorption, Distribution, Metabolism, and Elimination) attributes, and in vitro potency property space for a set of CNS drugs and candidates.1 In that work, the two compound sets differentiated by physicochemical property, ADME and safety attributes; we utilized the drug set to define a historical optimal chemical space possessing alignment of key drug properties for CNS therapeutic agents. In an effort to use this knowledge prospectively in design, before compounds are synthesized, we searched for a way to incorporate this information into an easy-to-use design tool. We began experimenting with the concept of multi-parameter optimization.

Multi-parameter optimization methods

provide a means to assess and balance several variables based on their importance to the overall objective.2 Using this approach we developed the CNS MPO desirability tool, which consisted

2 ACS Paragon Plus Environment

Page 3 of 34

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


of six fundamental physicochemical properties [(a) lipophilicity, calculated partition coefficient (ClogP); (b) calculated distribution coefficient at pH = 7.4 (ClogD); (c) molecular weight (MW); (d) topological polar surface area (TPSA); (e) number of hydrogen bond donors (HBD); and (f) most basic center (pKa)] that we determined to be important factors in alignment of ADME and safety drug attributes.1,3 In contrast to a quantitative structure–activity relationship (QSAR) or machine learning model,4,5 the CNS MPO desirability tool is built upon medicinal chemistry experience on ranges of desirable property space and simple piecewise linear transformational functions with values between 0 and 1 (defined as T0 for each property).3 A monotonic decreasing function was used for ClogP, ClogD, MW, pKa, and HBD, while a hump function was used to define TPSA, Figure 1.3 Each parameter was weighted equally and the collective score ranged from 0 to 6, with higher CNS MPO scores being more desirable. Most of the compounds in the drug set had CNS MPO desirability scores above 4, which differentiated them from the historical candidate set. The advantages of the CNS MPO desirability method lie in: a) its simplicity, with parameters derived from best medicinal chemistry practices; b) its ability to balance multiple variables while avoiding hard cutoffs; and c) its demonstrated alignment with desirable in vitro ADME attributes. Importantly, CNS MPO can be used prospectively in molecular design. Prior to the publication of our work in 2010, we began to routinely use the CNS MPO desirability method to help prioritize design ideas for advancement to synthesis. After nearly 8 years of use, we have re-assessed the robustness of this tool and report herein the analysis and comparison of our second wave of candidates, post-CNS MPO implementation, with the original candidate set.



ClogP

B) 1.0

0.8

0.8

0.6 5

0.4

3

0.4

-2

0

2 4 ClogP

6

8

MW

-6

0.8

0.8

T0 TPSA

1.0

500

0.6 360

0.4

-2

0 2 ClogD

4

6

8

0.6 20

0.4

120 90

40

0.2

0.2

0

0 200

300 400 MW

500

600

HBD

0

20 40 60

80 100 120 140 TPSA

pKa

F) 1.0

T0 pKa

1.0 0.8 0.6 0.4 0.2 0

-4

TPSA

D)

1.0

E)

2

0

0

C)

4

0.6

0.2

0.2

T0 MW

ClogD

1.0

T0 ClogD

T0 ClogP

A)

T0 HBD

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 34

4 0

0.8 0.6 8

0.4

10

0.2 0 0

1

2 3 HBD

4

5

2

4

6 pKa

8

10

12

Figure 1. "Reprinted with permission from (Wager, T. T., Hou, X., Verhoest, P. R., and Villalobos, A. (2010) Moving beyond rules: The development of a central nervous system multiparameter optimization (CNS MPO) approach to enable alignment of druglike properties. ACS Chem. Neurosci. 1, 435-449.). Copyright (2010) American Chemical Society." Each plot represents one of the six physicochemical property desirability functions used to generate the CNS MPO. Each point on a plot represents a drug or candidate. A) ClogP, B) ClogD, C) MW, D) TPSA, E) HBD, F) pKa. The most desirable (T0 = 1.0) and least desirable (T0 = 0.0) 4 ACS Paragon Plus Environment

Page 5 of 34

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


inflection points are marked with green and red arrows, respectively. A linear function was used to determine the desirability scores between the inflection points.3 Results and Discussion Twenty-one clinical candidates identified post CNS MPO implementation (“CNS MPO candidate set” from herein), and the original candidate (108) and drug (119) sets from our 2010 publication1, were evaluated using the original set of six physicochemical properties used to define the CNS MPO algorithm: ClogP, ClogD, MW, TPSA, HBD, and pKa (Figure 2). The ClogP median value (2.2) for the CNS MPO candidate set is a log unit lower than the previous candidates (3.3) and one-half log unit lower than the drug set (2.8). The ClogP values for the majority of the CNS MPO candidates varied from 1.7 (25th percentile) to 3.0 (75th percentile). The median value for ClogD (3.0) of the CNS MPO candidate set was higher than the corresponding ClogD median values for the drug and candidate sets (1.6 and 2.3, respectively). Comparison of the most basic pKa median value for the three sets of compounds showed that the CNS MPO candidate set had a lower median basic pKa (4.1) than the drug (8.0) and candidate (8.3) sets; this difference of 4 log units was statistically significant. Given that the CNS MPO candidate set is more neutral in nature, the higher median value for ClogD (3.0) vs. ClogP (2.2) for this compound set suggests that the ClogD calculator may be over-estimating ClogD values (vide infra). The CNS MPO candidate set MW (397.9) median value was higher than the drug set MW (305.3) and the candidate set MW (360.4) by 94.6 and 37.5 Da, respectively. Polarity, as described by topological polar surface area (TPSA), ranged from 73.4 Å2 (25th percentile) to 88.4 Å2 (75th percentile) with a median value of 80.8 Å2 for the CNS MPO candidate set, nearly double the drug set (44.8 Å2). The shift to more polar property space was intentional and designed to improve safety outcomes, this is discussed later in the article.

All three sets of



compounds had a minimal number of HBD, with the median value being one HBD, suggesting that optimization of HBD to ≤ 1 may increase the odds of identifying CNS-penetrant compounds. From this analysis, it is clear that the CNS MPO candidate set occupies a different property space from either the original candidate or drug sets. By focusing on the multiparameter approach, rather than being limited by single parameters and hard cutoffs in design, we have expanded “traditional CNS space” to include molecules that are less lipophilic, less basic, more polar, and larger while retaining good CNS exposure. This approach has enabled access to new CNS target classes such as kinases and proteases, which may require different physical chemical property ranges in order to achieve higher potency. ClogP

ClogD

MW

9 9 7

1

z

MW

ClogD

ClogP

500

5

3

3 1

100

-5 Candidates 108 2.2 3.3 5.1 3.4

CNS MPO Candidates 21 1.7 2.2 3.0 2.2

Drugs 119 1.6 2.8 4.0 2.7

Candidates Count 25% Median 75% Mean

108 0.4 2.3 3.3 2.0


TPSA

Drugs 119 0.7 1.6 2.5 1.6


120

4

60

z

3

pKa

HBD

80

2

z

1

20 0

0

Count 25% Median 75% Mean

108 39.6 51.2 65.5 53.7


Drugs 119 32.3 44.8 63.3 48.5


108 0 1 2 1.2

CNS MPO Candidates 21 0 1 1 1.0


Drugs 119 248.3 305.3 345.9 303.5

pK a

100

Candidates

108 307.5 360.4 419.6 357.4

HBD

5

140

40

z

200

-3

-3

400 300

z

-1

-1

Count 25% Median 75% Mean

600

7

5

TPSA

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 34

12 11 10 9 8 7 6 5 4 3 2 1

Drugs 119 0 1 2 1.1


108 7.2 8.3 9.2 7.6


Drugs 119 3.1 8.0 9.3 6.6

Figure 2. Physicochemical property distribution of drugs, candidates, and candidates post-CNS MPO implementation (CNS MPO candidates) for ClogP, ClogD, MW, TPSA, HBD, and most


Page 7 of 34

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


basic pKa. BioByte (version 4.3) was used to calculate ClogP and ACD software (version 12.1) was used to calculate ClogD and pKa. Count represents the number of compounds included in each analysis. Red dotted line represents median value. We calculated the CNS MPO desirability scores for the three compound sets and compared the compound distribution across the desirability continuum (0–6), Figure 3.3 The CNS MPO candidate set had a higher overall desirability score than the candidate and drug sets despite the fact that median ClogD and MW values were in the less desirable range. The improvement in CNS MPO desirability is achieved because the median ClogP and TPSA values were in a more desirable range, reinforcing the value of a flexible multi-parameter approach. The CNS MPO candidate set had the highest numerical percentage of compounds with CNS MPO desirability score > 5 (48%) compared to both the drug (40%) and candidate sets (30%). The original candidate set had a large percentage (31%) of compounds in the 3–4 range, while the CNS MPO candidate set had only 19%, which is comparable to the drug set (16%). The overall distribution of the CNS MPO candidate set resembles the drug set, with the number of compounds in each bin increasing with the desirability score. Further, no compound in the CNS MPO candidate set exhibited a desirability score less than 2. Collectively, this suggests that our compound designs are effectively using CNS MPO to explore and balance compound properties.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 34

Figure 3. CNS MPO desirability scores for candidates, CNS MPO candidates, and drugs were plotted from low to high CNS MPO desirability score along the x-axis. The compound count and percentage for each bin appear above the corresponding bar. In our original analysis of drugs and candidates, the probability of a compound possessing desirable in vitro ADME attributes (high Papp, low P-gp efflux liability, low unbound human liver microsome clearance) increased as its CNS MPO desirability score increased. In the current analysis, the CNS MPO candidate set had a higher CNS MPO median value (5.0) than the original candidate set, and indeed, across all three ADME assessments the CNS MPO candidate set provided a higher percentage of optimal values (Figure 4). All 21 compounds in the CNS MPO set were tested in each of the in vitro assays: 86% of the compounds had high passive permeability, 90% exhibited low P-pg efflux liability and 90% displayed low metabolic clearance. It is interesting to note that the CNS MPO candidate set also outperformed the drug set in having a higher percentage of compounds with optimal ADME values.


Page 9 of 34

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Candidates

CNS MPO Candidates

Drugs

21

78

82

A)

14%

22%

5% 17%

48% Papp Papp > 10 2.5 < Papp ≤ 10 Papp ≤ 2.5

30% 85

B)

78%

86% 10%

80

21 25%

45% P-gp ER ER ≤ 2.5 ER > 2.5

55% 75% 90% 93

C)

10%

90

21 30%

50% HLM CLint,u ≤ 100 CLint,u > 100

50% 70% 90%

Figure 4. Distribution of in vitro permeability Papp, P-gp efflux ratio, and unbound human liver microsome (HLM) intrinsic clearance (CLint,u), for candidates, CNS MPO candidates, and drugs. A) Binned values for Papp obtained from the RRCK assay, color-coded by high permeability (Papp > 10, green), moderate permeability (2.5 < Papp < 10, yellow), and low permeability (Papp < 2.5, red) in units of 10-6 cm/sec. B) Binned values for P-gp efflux liability obtained from the MDCKMDR1 assay, color-coded by low P-gp liability (ER ≤ 2.5, green), or high P-gp liability (ER > 2.5, red). C) Binned values for clearance (CLint,u) assessed in a human liver microsome stability assay, color-coded by low clearance (CLint,u ≤ 100 mL/min/kg, green) and high clearance (CLint,u > 100 mL/min/kg, red. Pie charts are color-coded based on the value of the bin, from desirable values (green) to undesirable values (red), and the number of compounds in each pie is shown above each pie graph. 9 ACS Paragon Plus Environment


Previously, we reported that an increase in CNS MPO desirability score led to an increase in the number of individual compounds with alignment of in vitro ADME attributes defined as a compound satisfying one or more criteria of high Papp, low P-gp ER, and low CLint,u. We compared the three sets to determine if the ADME alignment trend would hold true for the new compound set (Figure 5). A clear distribution of compounds exhibiting varying degrees of alignment was observed across the CNS MPO desirability continuum. Compounds having full alignment of ADME attributes heavily populated the higher end of the CNS MPO score spectrum; for the CNS MPO candidate set, the preponderance (94%) of compounds with CNS MPO scores above 4 displayed full alignment of ADME attributes.

50

45

40

Candidates ADME Attributes

CNS MPO Candidates

Drugs

3 2 1 0

48 17 44 19

35

Count

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 34

33 8

33 14

30

32 3

25

15

25

17

3 4

20 7 15 11

10

2 9

7 3 5

8 10 9 4

4

3 3

0

19 10

5

x≤2

2

Central Nervous System Multiparameter Optimization Desirability

Recommend Documents