1 - Free

advertisement
A BIT OF ADME FOR EVERYONE:
LogD, SOLUBILITY, PERMEABILITY
AND
THE UNIVERSAL DETECTION SYSTEM
1
CONFIDENTIAL
OUTLINE
• ADMET in modern discovery
• A brief refresher:
– LogD, Solubility, Permeability
– Issues, methods, tradeoffs
• More interesting ADME assays:
– Caco/PAMPA-less BBB permeability
• The Universal Detection System - UDS
2
CONFIDENTIAL
The Miracle of a Drug
Stomach Intestine
Solubility
pH 3-8
Stability
 pH 2
 enzymatic
Stability
Phase I4
Portal Liver
Vein
Stability1
 Phase I4
 Phase II
Blood
Protein
Binding4
Stability
 pH 3-8
 enzymatic
Urinary
Clearance
2
3
4
3
Inducible
Active transport increases permeability of some compounds
Intestinal re-uptake of some compounds
Drug-drug interaction potential
CONFIDENTIAL
Cell
Distribution
Cellular
(whole body)Partitioning
Stability
Enzymatic
Permeability2 Biliary
3
(epithelium) Clearance
Passive
 pH 3-8
 Pgp Efflux4
1
Organ
Kidney
Permeability2 Permeability2
 Passive
 Passive
4
 Pgp Efflux  Pgp Efflux4
(Organ
Specific)
After Susan Petusky
Wyeth Research
ADME in Discovery: Many Strategies (1)
Sequential

Purity
•Data integrity
• Compound
conservation

LogD
Solubility
• Too slow

Permeability
4
CONFIDENTIAL

ADME in Discovery: Many Strategies (2)
Weighted Score
5
Purity
85%

LogD
2.2
0.85
Solubility
10

Permeability
10-5
Metabolism
Protein binding
CONFIDENTIAL
• Holistic
4
3.4
0.5
2
1.0
30%
0.75
5
3.75
99%
0.9
1
0.9
9.05
LogP/LogD (1)
• Equilibrium partition coefficient between
1-octanol/buffer
• LogP – non-ionized compound
• LogD – ionized compound – f(pH)
• Good Absorption characteristics:
– LogP ~ 2.5
6
CONFIDENTIAL
Lipophilicity in Absorption
Navia MA, et.al., DDT 1,
(5) May 1996
7
CONFIDENTIAL
LogP/LogD (2)
• LogD > 5:
–
–
–
–
Tough to measure
Promiscuous binders
Poor solubility, oral absorption
Strong CYP450 interaction
• LogD 0-3:
– Best balance of solubility, permeability
• LogD < 0:
– Good solubility, poor permeability
8
CONFIDENTIAL
Measuring LogD
• The real thing: Shake flask method
• RP HPLC:
– See K. Valko, J. Chrom. Sci, 1037 (2004)
• pH-metric
• Microemulsion electrokinetic
chromatography
• Calculating:
– ACD LogP
– CLogP
– PrologD
9
CONFIDENTIAL
A Brief Reminder: LogD is Difficult
Buffer Composition
pH
LogD
0.01 M Universal Buffer1
0.01 M K-Phosphate Buffer
0.01 M Na-Phosphate Buffer + 0.15 M NaCl
0.01 M Universal Buffer1 + 0.15 M NaCl
0.10 M K-Phosphate Buffer
0.01 M Universal Buffer1
0.10 M K-Phosphate Buffer
0.07 M Na-Bicarbonate Buffer
7.4
7.4
7.4
7.4
7.4
11.0
11.0
11.0
0.015  0.009
0.010  0.007
0.632  0.009
0.697  0.011
0.981  0.008
1.147  0.004
1.530  0.002
2.440  0.010
Compound: Propranolol
1 – Universal buffer is composed of a mixture of acetic,
phosphoric, and boric acids with NaOH
10
CONFIDENTIAL
ADW:
LogD Assay
11
CONFIDENTIAL
Validation: LogD Accuracy
6
logD
ADW
4
2
0
-2
-4
-4
-2
0
2
4
6
manual
logD
LogDADW = 0.002(± 0.008) + 1.011(± 0.005)*LogDmanual
N = 179; r2 = 0.9960; standard error of estimate =
0.1022
12
CONFIDENTIAL
A Simple Concept
• Solubility = Concentration of a
dissolved compound in equilibrium with
its solid
• But:
– Which solid?
• Equilibrium (most stable form) vs. apparent
(other forms)
– Which solvent?
• Buffers (intrinsic?) and co-solvents (kinetic)
– Which equilibrium?
• Time (kinetic) and temperature
13
CONFIDENTIAL
Key Differentiating Factor
• Are you measuring
the actual
concentration?
Elemental:
Yes, directly
Elemental Concentration:
Carbon
Nitrogen
...
No standards / calibration
Evaporative Light
Scattering
No standards / calibration
Absorbance:
Yes, indirectly
Static Light Scattering
Turbidity:
No, solubility is
inferred from dilution
factor off a standard
14
CONFIDENTIAL
Compound specific
Absorbance
Can You Tell The Saturated Solution?
Saturated solutions
Phosphate buffer, pH 11
1 week incubation
Left to Right:
ChlorpromazineHCl
Bendroflumethiazide
Clofazimine
Bifonazole
ThioridazineHCl
TriflupromazineHCl
Nifedipine
Perphenazine
PromazineHCl
15
CONFIDENTIAL
Assay Effects: Particles
• Solid particles are an integral part of
the solubility assay
– Particles are always present
– They must be present for turbidity to work
– They are artifacts in absorbance/elemental
assays
• The effects of particles on the data
must be considered when examining
the data
– Subtle to substantial influence on quality of
results
16
CONFIDENTIAL
Permeability: The Real Thing
Lipid bilayer (10,000,000 X)
“The Machinery of Life”
David Goodsell
Copernicus (Springer-Verlag)
17
CONFIDENTIAL
Gap junction (1,000,000 X)
Two Choices
Measure
average
concentrations
Calculate
permeability

Bilayer chemistry
 Transport physics
 Experimental details


18
Experimental
database
Structural
parameters
CONFIDENTIAL
Establish
relationship
Measure
structural
parameters
Plug into
relationship
Calculate
permeability
PAMPA: Stirring + Acceptor Sink
Drug
19
CONFIDENTIAL
Scavenger
Drug + Scavenger
BBB Permeability:
An Alternative Approach*
• Instead of:
– Trying to find the “ultimate” membrane model
– Deciphering permeability from a complex
experiment
• Use real-world in vivo data + structural
descriptors:
– What prediction level is necessary?
– Which descriptors are useful?
*Gulyaeva et. al, EJMC 38 (2003)
21
CONFIDENTIAL
Data for BBB Permeation
#
CNS
Compound
#
CNS
Compound
#
CNS
Compound
1
-
Acyclovir
23
-
Tiapride
44
+
Mequitazine HCl
2
-
Acetaminophen
24
+
Amitriptiline HCl
45
+
Minaprine
3
-
Albendazole
25
+
Chlorpromazine HCl
46
+
Mefexamide
4
-
Albuterol
26
+
Clomipramine
47
+
Naltrexone HCl
5
-
Ampicillin
27
+
Clonidine HCl
48
+
Naloxone
6
-
Antipyrine
28
+
Desipramine HCl
49
+
Nortriptyline
7
-
Astemizole 2HCl
29
+
Doxepin
50
+
Perphenazine 2HCl
8
-
Atenolol
30
+
51
+
Physostigmine
9
-
Cimetidine
Doxylamine
succinate
52
+
Progesterone
31
+
Estrone
53
+
Promazine HCl
32
+
Fluoxetine HCl
54
+
Promethazine HCl
33
+
Flupentixol 2HCl
55
+
Propranolol HCl
34
+
Fluphenazine
56
+
Protriptyline
35
+
Haloperidole HCl
57
+
Pyrilamine
36
+
Homochlorcyclizine
58
+
Thioridazine HCl
37
+
Hydroxyzine 2HCl
59
+
Tranylcypromine
38
+
Ibuprofen
60
+
Trazodone
39
+
Imipramine HCl
61
+
Trifluoperazine
40
+
Indomethacin
62
+
Triflupromazine
41
+
Lidocaine HCl
63
+
Trimipramine
42
+*
10
11
12
13
14
15
22
+: logBB > 0.3
--: logBB < -1.0
-*
-
Domperidone HCl
Ebastine
5-Fluorouracil
Ftorafur
Furosemide
5Hydroxytryptophan
e
16
-
Iproniazid
17
-
Metoclopramide
18
-
Metoprolol
19
-
Metronidazole
20
-
Phenelzine
21
-
Pirenzepine HCl
CONFIDENTIAL
Loperamide HCl
Structural Descriptors (1)
• Basic Premise:
– Only structural differences matter – since
compounds are predominantly different by
structure
• Key Methodology:
– Ask for minimally acceptable answer:
• Must you know a number or will a classification
suffice?
– Use smallest number of descriptors:
• Principle of parsimony
23
CONFIDENTIAL
Structural Descriptors (2)
• LogD(7.4):
– Measure of relative affinity between polar
and non-polar media
– Obtained from experiments or calculations
• N(CH2):
– Measure of relative affinity between two
aqueous solvents of different structures
– Obtained using aqueous two-phase
partitioning experiments (Dex-PEG
systems)
24
CONFIDENTIAL
Predictive Equation
• Predict what?
– Probability of a compound being CNS+
• Predict how?
– Logistic regression model using penalized maximum
likelihood (nested models) or information criteria
(non-nested models)
A = - 15.04 + 12.56*log D - 0.4* LogD * N (CH 2 )
P(CNS + ) = exp( A) /(1 + exp( A))
• Predictive accuracy:
– 96.6% using cross validation procedures (one point
out of the model for all 63 data points)
25
CONFIDENTIAL
P(CNS+)
120
100
P(CNS+)
80
60
40
20
5
25
20
1
15
10
N(CH
2)
26
CONFIDENTIAL
0
5
0
Lo
g
2
-20
.4
D(
7
3
)
4
0
Typical Big Pharma - ANALIZA
Collaboration
• LogD (pH =7.4)
• Solubility (pH = 6.5)
• Thermodynamic solubility (dry)
• Over 21,000 data points in 2005
• Real data: high throughput,
miniaturized shake flask methods
27
CONFIDENTIAL
1 Year Precision Data: Solubility Controls
harmine
imipramine
sulfamethizole
600
500
uM
400
300
200
100
0
12/29/2004
2/17/2005
4/8/2005
5/28/2005
7/17/2005
Assay Date
28
CONFIDENTIAL
9/5/2005
10/25/2005
12/14/2005
1 Year Precision: LogD Controls
harmine
imipramine
sulfamethizole
4
3
log D
2
1
0
1/3/2005
2/22/2005
-1
-2
-3
29
CONFIDENTIAL
4/13/2005
6/2/2005
7/22/2005
9/10/2005
10/30/2005
12/19/2005
1 Year Statistics
LogD
Solubility
Plate
Total
Aggreg.
rate, %
Total failure
rate, %
Net failure
rate, %
Net failure
rate, %
A
3,811
5.8
6.1
0.3
0.7
B
3,781
6.4
6.9
0.5
0.7
C
2,758
2.5
3.0
0.5
D
2,833
5.6
6.3
0.7
30
CONFIDENTIAL
Universal Detection System – UDS
Overview
31
CONFIDENTIAL
The Need
• We have automated ID solutions: MS, etc.
• Pharmaceutical compounds in
discovery/development are impure:
– Combichem libraries are 90-95% pure
– Compounds synthesized during LD are 90+% pure
– Serious implications for:
• Early ADME evaluation (e.g., solubility)
• Late selection and lead optimization
• Process development and FDA submissions
• Impurity quantification is a vexing problem:
– Impurity has to be separated
– Standards must be prepared
– Very difficult in practice!
• Highly desirable:
– Automated quantification method for all impurities in
a sample w/o standards
32
CONFIDENTIAL
Unique Advantages
• Orthogonal detection methodology to Ultra Violet
(UV) or Mass Spectrometry (MS)
• Equimolar detection for instrument universal
calibration curve
• Wide applicability to most pharmaceutical
compounds (93% +)
• Large intrinsic dynamic range (ca. 0.1-50,000 ppm
nitrogen)
• Complete automation:
– On-board separation capabilities with individual peak
concentrations
– Peak zoom capability to focus on assaying of minor
peaks
– Automated optimization with variable instrument
gain and injection volume
33
CONFIDENTIAL
UDS Architecture
Other Agilent
1100 modules
PDA
data
Agilent 1100
PDA
Sample
Antek 8060
Agilent 35900E
ADC
Analiza CTRL
N Gain, AZ
Analiza
NDCS
PC
Agilent
ChemStation
34
CONFIDENTIAL
Development System – Hardware
35
CONFIDENTIAL
Development System – SW
36
CONFIDENTIAL
UDS Operating Modes
• Calibration
– Construct universal piece-wise linear curve
• Quantification
–
–
–
–
–
37
Noise A(gressive)
Noice C(onservative)
P(arent) P(eak) G(ain)
P(arent) P(eak) A(rea)
O(ptimize) V(olume)
CONFIDENTIAL
UDS Sequence Table – After Run
38
CONFIDENTIAL
Universal Calibration Example
Area Under The Curve
mg/ml
ppmN
Inj. #1
Inj. #2
Inj. #3
Avg.
1
0.0005
0.144
113
114
96
108
2
0.001
0.289
198
163
171
177
3
0.0025
0.721
389
399
393
394
4
0.005
1.443
802
803
786
797
5
0.01
2.886
1617
1629
1608
1618
6
0.025
7.214
4284
4283
4255
4274
7
0.05
14.428
8871
8818
8866
8852
8
0.1
28.856
18549
18502
18340
18463
9
0.25
72.141
47908
47484
47484
47625
10
0.5
144.281
97254
96869
95795
96639
11
1
288.563
193570
193361
193109
193347
2.5
721.407
460138
462725
466960
463274
13
5
1442.814
849204
841916
852793
847971
14
10
2885.627
1494800
1486750
1476160
1485903
CONFIDENTIAL
1e+6
1e+5
1e+4
1e+3
1e+2
1e+1
12
39
1e+7
Area Under the Curve
#
0.1
1
10
100
ppm Nitrogen
1000
10000
Universal Calibration – UDS Screen
40
CONFIDENTIAL
UDS Performance Tests
Sample
S
AC
0.025:0.05
A
AC 0.01:1
AC 0.1:10
mg/ml
ppmN
Inj.
#1
Inj.
#2
0.025
2.3163
2.269
2.267
2.31
2.284
2.28
-1.5
C
0.05
14.4281
14.79
14.78
14.89
14.77
14.81
2.6
A
0.0083
0.7721
0.723
0.668
0.712
0.703
0.70
-9.1
C
0.8333
240.47
240.7
238.25
239.32
238
239.07
-0.6
A
0.09804
9.0835
9.362
9.358
9.265
9.26
9.31
2.5
C
9.8039
2829.04
2881.7
2892.7
2889.8
2869.6
2883.4
1.9
A – Acetaminophene
41
CONFIDENTIAL
C - Caffeine
Inj.
#3
Inj.
#4
Avg.

%
UDS Report Screen
42
CONFIDENTIAL
Thank You
43
CONFIDENTIAL
Download