Biomimetic HPLC methods to

predict in vivo drug distribution

and improve candidate quality

Klara Valko

Analytical Chemistry

MDR Chemical Sciences

GlaxoSmithKline

Stevenage, UK

Problems causing attrition of

drug discovery compounds

Dose

Solubility

Permeability

Free

Concentration

Toxicity

Target

Silent binding sites

Non-specific binding to proteins

and phospholipids

Elimination

Metabolism and clearance

Poor

absorption

Poor

efficacy

Poor safety

Which physchem parameters

influence attrition?

Poor

absorption

Poor

efficacy

Poor safety

Low solubility, poor permeability

Low free concentration at the site

of action due to high protein and

phospholipid binding

Due to high lipophilicity, high

tissue partition, promiscuity, etc

How can chromatography (HPLC) help?

HPLC retention can be related to

association constants between the solute

and the stationary phase

Cs. Horváth, W. Melander, I. Molnár, J. Chromatogr. 125 (1976) 129.

Measuring bio-relevant association constants has

great impact on drug discovery!

Based on the solvophobic theory the interaction between the solute

and the stationary phase is considered as a reversible association of

the solute molecules with the stationary phase moiety

(hydrocarboneous, membrane, or protein). Accordingly solute

retention is governed by the dynamic equilibrium constant.

Vs = volume of the

stationary phase

Vm = volume of the mobile phase

Advantages of

chromatographic methods

A compound’s retention time can be directly related

to its distribution between the stationary and the

mobile phase, there is no need for concentration

determination.

By changing the stationary phase and the mobile

phase composition various types of lipophilic

interactions can be investigated.

Impurities do not disturb the measurements as they

are separated from the main peak. Measurements

can be completely automated!

Estimating lipophilicity using isocratic

retention factor at various mobile phase

composition

% organic phase in the mobile phase

log

k

10 20 30 40 50 60 70 80 90 100

0

-0.2

0.2

0.4

0.6

log k o = extrapolated log

k to 0% organic phase

φ0 is the organic

phase concentration

when log k = 0

φ0 φ0 φ0 φ0 φ0 φ0

1 2

3

4

5 6

K. Valko, P. Slegel;

New chromatographic

hydrophobicity index

(φ0) based on the

slope and the intercept

of the log k′ versus

organic phase

concentration plot, J.

Chromatogr. A. 1993

(631) 49-61.

Generic gradient reverse phase

chromatography - dramatic reduction of

analysis time without losing resolution

40 min

4 min

Reproduced from Separation methods in Drug Synthesis and Purification Ed.

K. Valko, Elsevier, 2000, Iain Mutton: Fast generic HPLC methods p. 72

(tg)

A B

CHI = A*tg + B

Chromatographic Hydrophobicity Index (CHI)

is proportional to the gradient retention time

Compound CHI7.4

at pH 7.4

CHI2

at pH 2

CHI10.5

at pH 10.5

Theophylline 18.4 17.9 5.0

Phenyltetrazole 23.6 42.2 16.0

Benzimidazole 34.3 6.3 30.6

Colchicine 43.9 43.9 43.9

Phenyltheophylline 51.7 51.7 51.7

Acetophenone 64.1 64.1 64.1

Indole 72.1 72.1 72.1

Propiophenone 77.4 77.4 77.4

Butyrophenone 87.3 87.3 87.3

Valerophenone 96.4 96.4 96.4

Calibration of CHI at pH7.4

y = 54.329x - 71.702

R2 = 0.9972

0.00

20.00

40.00

60.00

80.00

100.00

120.00

1.4 1.9 2.4 2.9 3.4

Chromatographic Hydrophobicity Index

4-minute method! LunaC18(2) 50 x 3 mm; 1.00 ml/min; Mobile phase A 50 mM ammonium acetate pH 7.4 and B is 100% acetonitrile.

Gradient: 0 - 2.5 min 0 - to 100% B; 2.5 - 2.7 min 100% B.

K. Valkó et al. Curr. Med. Chem. 8 (2001) 1137-1146

Ultrahigh Performance Liquid

Chromatography (uPLC)

CHI TM1

Time0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80

AU

0.0

5.0e-2

1.0e-1

1.5e-1

2.0e-1

2.5e-1

3.0e-1

CHI TM1_LunapH74 Diode Array 254

Range: 3.969e-1

1.02

0.86

0.64

0.590.72

0.92

1.27

1.16

1.10

1.37

The CHI test mix is separated in less than 90 sec

Now we can determine a compound’s lipophilicity in 90 sec using

various starting mobile phase pH Courtesy of Shenaz Nunhuck

Parallel measurement of a compound’s retention at

various pH to reveal acid/base character

Typical 4-way chromatograms of a base

pH2

pH7.4

pH 10.5

IAM 7.4

CHIs measured at 3 pHs provide an automatic way of grouping molecules

according to acid/base character without structural information.

0

10

20

30

40

50

60

70

80

90

100

Neutral

(Zwitterionic)

Strong acid Weak acid Strong base Weak base Amphoteric

pH2

pH7.4

pH10.5

The change of CHI values by changing the pHCHI

CHI values at pH 2, pH 7.4 and pH 10.5 reveal

acid/base character of compounds

Biomimetic HPLC measurement of Human

Serum Albumin (HSA), α-1acidglycoprotein

(AGP) and Immobilized Artificial Membrane

(IAM) partition

HSA AGP IAM

pH 7.4 aqueous mobile phases

IPA IPA ACN

Stationary phases

Biomimetic lipophilicity measurements

(Membrane partition) using Immobilised Artificial

Membrane stationary phase

Stationary phase IAM calibration y = 26.647x - 37.653

R2 = 0.9966

0

10

20

30

40

50

60

1 1.5 2 2.5 3 3.5gtR

CHI

Typical calibration

Compound gtR CHI IAM

Octanophenone 3.269 49.4

Heptanophenone 3.145 45.7

Hexanophenone 3.001 41.8

Valerophenone 2.822 37.3

Butyrophenone 2.601 32

Propiophenone 2.341 25.9

Acetophenone 2.013 17.2

Acetanilide 1.83 11.5

Paracetamol 1.591 2.9

Column: IAM PC2 (CH2)12 150 x 4.6

Mobile Phase flow rate: 2 ml/min

Gradient: 0 to 3 min 0 to 80% acetonitrile

3 to 3.5 min 80% acetonitrile

3.5 to 3.7 min 0% acetonitrile

Cycle time: 5 min

K. Valko et al. J. Pharm. Sci.89 (2000) 1085-1096

Calibration ploty = 2.177x + 0.1304

R2 = 0.9612

-0.50

0.00

0.50

1.00

1.50

2.00

2.50

-0.40 -0.20 0.00 0.20 0.40 0.60 0.80 1.00

logtR

logK

litera

ture

Calculate %Binding

logK = slope * log(tR) + int

K = %B / (101-%B)

y = 0.9309x - 0.3329

R2 = 0.879

0

20

40

60

80

100

0 20 40 60 80 100

HSA Column

Lite

ratu

re %

bin

ding

n=71

K. Valko et al. J. Pharm. Sci. 92 (2003) 2236

Column: HSA 50 x 3 mm (Chrom Tech, Chiral Technologies)

Flow rate: 1.8 ml/min at 300C

Mobile phase: 50 mM ammonium acetate pH7.4

Gradient: 0 - 3 min 0 to 30% 2-propanol;

3 to 10 min 30% 2-propanol;

10 to 10.5 min 0% 2-propanol

Cycle time: 15 min

Serum albumin binding measurement using

chemically bonded serum albumin stationary

phases

Retention time of compounds can be

converted to % binding or K association

constant

)(logexplog HSAKHSAkHSAke

2 min

AGP binding measurement by

HPLC

Same principle as HSA

binding measurements:

AGP column

2-propanol gradient

pH 7.4 ammonium acetate

Calibration with AGP binding

data derived from published

% AGP bound values

Typical AGP column calibration ploty = 2.7976x - 0.5289

R2 = 0.9744

-0.400

-0.200

0.000

0.200

0.400

0.600

0.800

1.000

1.200

1.400

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70

log tR

logK

AG

P

Calibration set of compounds: Nizatidine,

Bromazepam, Warfarin, Propranolol, Imipramin,

Nicardipine, Chlorpromazine

Log kplasma= 0.87*log k(HSA) +

0.17*logk(AGP) + 0.06 cMR-0.27

n=60 r=0.96 s= 0.44 F=240

cMR is the calculated molar refraction (size

parameter) that accounts for general lipophilic

binding of compounds to neutral plasma

proteins such as immunoglobulin.

Model compounds were chosen to have a wide

diversity in terms of charge, and no inter-

correlation between HSA and AGP binding.

Model for plasma protein binding

from HSA and AGP binding data Scatter Plot

calcKPPB

Lo

gk(p

las

ma

) fr

om

lit

era

ture

Logk(plasma) calculated from

measured HSA and AGP binding

Positively charged

compound

Negatively charged

compound

Neutral compound

Solvation process

Solute

molecule

Solvent molecules

(water)

First we need

to create a

cavity

Requires energy that

depends on the size

of the solute

Size descriptor: V

Solvation process Second step to build

new interactions

between the solute and

solvent molecules

We gain back

energy.

Solvent molecules

(water)

Solute

molecule

Solute –solvent

interactions

Dipole –dipole S

- - - +

+ -

+ -

+

-H ‐‐‐ O=C-

A B

Excess molar refraction

E

Molecular descriptors describe various

solvation and partition processes

SP= c + e* E + s*S + a*A +b*B + v*V

SP is the solute property in a given system

Molecular descriptors

Solvent

characteristics

M. H. Abraham, Scales of solute hydrogen bonding: Their construction and application

to physicochemical and biochemical processes. Chem. Soc. Rev. 22 (1993) 78-83.

Solvation equations for

biomimetic distributions

e/v s/v a/v b/v

Log K (HSA) 0.02 -0.07 0.16 -1.21

Log K (AGP) 0.46 -0.38 -0.33 -0.85

CHIRP,AcN 0.09 -0.24 -0.30 -0.98

logPoctanol 0.15 -0.28 0.01 -0.91

Log K (IAM) 0.11 -0.03 0.01 -1.05

Blood/brain 0.59 -1.03 -0.84 -0.78

water/skin 0.00 -0.33 -0.35 -1.95

•The octanol/water partition system is an excellent model for compounds binding to

HSA and IAM.

•AGP binding , CHI (reversed phase with acetonitrile) and blood/brain barrier partition

are different from octanol/water.

•Equations are derived from the data of non-ionized compounds.

IAM binds positively charged compounds

HSA binds negatively charged compounds

Scatter Plot

elogKHSA

e^log k(HSA)

e^

log

k(I

AM

)

Positively charged

compound

Negatively charged

compound

Neutral compound

Possible cases of poor

absorption

High lipophilicity correlates with low

solubility

ChromlogD + number of aromatic

rings provides solubility forecast

index

Permeability shows a parabolic

relationship with lipophilicity

Poor

absorption

ChromlogD + number of aromatic

rings forecast solubility issues

Binned ChromlogD pH7.4

-1 < x ≤ 0 0 < x ≤ 1 1 < x ≤ 2 2 < x ≤ 3 3 < x ≤ 4 4 < x ≤ 5 5 < x ≤ 6 6 < x ≤ 7 7 < x ≤ 8 8 < x ≤ 9 9 < x ≤ ... 10 < x

0

1

2

3

4

5

6

7

Measured CLND solubility < 30 M; 30-200 M >200 M

A. P. Hill, R. J. Young, Getting physical in drug discovery; a contemporary perspective on

lipophilicity and solubility. Drug Discovery Today, 15 (2010) 648.

Lipophilicity, permeability and

solubility A

rtific

ial M

em

bra

ne p

erm

eabili

ty n

m/s

ec

ChromlogD at pH 7.4 (obtained on C-18)

•Optimum lipophilicity is

needed for maximum

permeability

•Too lipophilic

compounds are less

soluble, may partition

into the membrane but

they have reduced

permeability

Modified Absorption Potential

(MAP)

Permeability is replaced by octanol/water partition coefficient (Kow)

and solubility is expressed by aqueous solubility of the non-ionized

form (i. e. intrinsic solubility) (Sw), D is the dose, number 4 came out

as an estimate of lumenal volume (0,25 L). The fraction of non-

ionized form is not needed, as ionization effects solubility and

lipophilicity in an opposite way, so it cancels out in the product term.

Ni, N., Sanghvi, T., Yalkowski, S. H.

(2002) Independence of the product term

of solubility and distribution coefficient of

pH. Pharm. Res., 19, 1862.

Compartment model and “free

drug hypothesis”

Free drug concentration should be the same (unless permeability barrier or active transport

distorts it)

Can be measured

by plasma protein

binding

Can be measured

by tissue binding

Volume of

distribution

free bound

Unbound volume of distribution/drug efficiency max

HSA

binding HSA + IAM

binding

Poor

efficacy

Estimating a compound’s distribution in

vivo from bio-mimetic HPLC measurements

Scatter Plot

predlogVd

logVd= 0.131elogK(IAM) - 0.2672*elogK(HSA) -0.305

n=130 r=0.88 s=0.31

Hollosy, F., Valko, K., Hersey, A., Nunhuck, S., Keri, Gy.,

Bevan, C., Estimation of volume of distribution in humans

from high throughput HPLC-based measurements of

human serum albumin and immobilized artificial

membrane partitioning. J. Med. Chem. 49 (2006) 6958.

The model works for acidic, neutral

and basic compounds. Compound’s

binding to phospholipids relative to

plasma proteins drives compound to

the tissue compartment from the

plasma.

Positively charge compounds bind

strongly to IAM, while negatively

charged compounds bind strongly to

HSA.

In v

ivo

lo

gV

d

Estimated logVd

Bio-mimetic HPLC data to explain in

vivo drug distribution

Though both protein binding and phospholipid

binding are driven by lipophilicity, the presence of

positive or negative charge drives the compound to

the tissue or to the plasma compartment.

These differences between the positive and

negative charges do not manifest in octanol/water

lipophilicity

logD drops whether the charge is positive or

negative

Both HSA binding and phospholipid binding are

affected by the shape of the molecules too.

Examples of structural changes

affecting volume of distribution

Nifedipine Amlodipine

Vd= 0.7 L/kg Vd= 21 L/kg

t1/2= 6 h t1/2= 34 h

HSA% = 88.8% HSA%= 91.4%

CHI IAM = 30.4 CHI IAM= 51.0

Predicted Vd= 0.7 L/kg Predicted Vd = 12.2 L/kg

Calc logD=logP=3.58 calc logD= 1.50 calclogP=3.01

No charge at pH 7.4 basic pKa = 9.45

The model from HSA and CHI IAM predicts the big differences in volume of

distribution.

Drug Efficiency - Definition

DOSE

EFFECT

F% Abs

Vd Cl

BTB PBB

Pgp Perm.

MoA pKi

[C]

[c]

EFFIC

IEN

CY

A drug is more efficient when a higher

concentration can be found at the site

of action with the smallest possible

dose.

x100(µg/g)Dose

(µg/mL) ConcBiophase DRUGEff

In-vitro efficiency ≈ 100%

In-vivo efficiency << 100%

S. Braggio, D. Montanari, et al. Expert Opin Drug Discovery,

Drug efficiency: a new concept to guide lead optimization

programs towards the selection of better clinical candidates. 5

(7) (2010) 609

Maximum drug efficiency (Drugeff max)

When the dose is absorbed 100% and distributed in the

plasma and tissue compartments without active transport or

permeability barrier.

Drugeff max can be estimated from the dose and the free

plasma concentration

The maximum drug efficiency can be achieved when the free

plasma and free tissue concentrations are the same

The fraction unbound in plasma and tissue may be very

different

log(DRUGeff max HPLC)

-1.5 -1 -0.5 0 0.5 1 1.5 2

-1.5

-1

-0.5

0

0.5

1

1.5

2

2.5

In Vivo DRUGeff max vs HPLC DRUGeff max

XY 0231.10423.0

81.02

R

Log HPLC DRUGeff max

=2.72-0.23*logKHSA-0.43*logKIAM

n=70 r2=0.81 s=0.32 F=179

Log D

RU

GE

ff M

ax

K. L. Valko, S. B. Nunhuck, A. P.

Hill, J. Pharm. Sci. 100, (2011) 849.

Influence of HSA binding on the

DRUGeff max

Strong albumin binding does not necessarily mean low drug efficiency

Log H

PL

C D

RU

GE

ff M

ax

Log k HPLC HSA

Colour: by

target class

Size by: dose

K. Valko, E. Chiarparin, S. Nunhuck, D. Montanari, In vitro measurements

of drug efficiency index to aid early lead optimization. J. Pharm. Sci. 101

(2012) 4155

Influence of phospholipid partition

on DRUGeff max

Phospholipid binding has high impact on DRUGeff max

IAM= Immobilised Artificial Membrane

Log H

PLC

DR

UG

Eff M

ax

Log HPLC IAM

Scatter Plot

clogp_day.value

-2 0 2 4 6 8 10

-1.5

-1

-0.5

0

0.5

1

1.5

2

DRUGeff vs clogP

Log DRUGeff inversely relates to clogP

lo

gD

RU

GE

ff

clogP

Size according to the dose

7TM

Enzyme

Ion Channels

Nuclear

Receptors

Undefined

K. Valko, E. Chiarparin, S. Nunhuck, D. Montanari, In vitro measurements

of drug efficiency index to aid early lead optimization. J. Pharm. Sci. 101

(2012) 4155

Possible cause of poor

efficacy

Poor

efficacy Strong IAM binding, indicating strong

tissue binding

Strong HSA binding indicating high

plasma protein binding

Reduced free concentration available at

the target relative to dose

Low drug efficiency (DRUG eff max)

Possible cause for side

effect/toxicity

Poor safety Strong IAM binding, indicating strong

tissue binding

Poor drug efficiency (indicated by strong

HSA and IAM binding) which means

higher dose is needed to achieve

efficacious free concentration at the site

of action

High dose increase the chance of safety

issues

Positive charge and strong phospholipid

binding indicates phospholipidotic

potential

Casartelli, et al. A cell-based approach for the early assessment of the phospholipidogenic

potential in pharmaceutical research and drug development. Cell biology and Toxicology,

19 (2003) 1610176.

Conclusions We can obtain molecular physicochemical properties of drug discovery

compounds by HPLC retention time measurement.

The gradient retention times can be converted to binding constants

using the data of a calibration set of compounds.

Retention on C-18 stationary phase can be used to estimate compound’s

partition into hydrocarbons. It is inversely related to solubility and shows

parabolic relationship to permeability.

Retention on albumin (HSA), α-1-acid glycoprotein (AGP) and

immobilised artificial membrane (IAM) stationary phase can be used to

estimate volume of distribution and drug efficiency.

This parameters can be related to absorption, distribution and safety of

discovery compounds and can help the lead optimisation process to

select drug candidates that are less likely to fail during development

Acknowledgements

Alan Hill, Shenaz Nunhuck, Iain Reed and Silvia

Bardoni (GSK Chemical Sciences-Analytical

Chemistry Physicochemical Characterisation

Group)

Simon Readshaw, Bob Boughtflower (GSK

Chemical Sciences-Analytical Chemistry)

Chemistry and DMPK teams working on various

programs

Co-authors of publications