Dressman APV 2013

Successful early stage development of

pharmaceuticals: Preclinical in vitro testing: bridging the testing of

formulations in preclinical development to the clinic using biorelevant media

APV, Darmstadt

05.-06. März 2013

Prof. Dr. Jennifer Dressman

Dressman

Model of Oral Dosage Form Performance

Dosage

Form Drug in

Solution

Blood

Site of

Action

Therapeutic

Effect

In-vivo

Dosage Form

Performance

Pharmacokinetic

Measurement

Clinical / PD

Measurement

When can in-vitro performance testing serve as a surrogate for in-vivo performance? G

ut W

all

In-vitro

Dissolution

Solubility Permeability

Based on a slide from 2007 AAPS-FDA BCS, BE, and Beyond Workshop Presentation, entitled General BA/BE Issues, Dale Conner,

Division of Bioequivalence, Office of Generic Drugs, CDER, FDA APV 2013

Linking the lab to the patient……

Old paradigm: many drug candidates were

highly soluble and simple solubility and dissolution tests sufficed to guarantee performance

New Paradigm: many drug candidates are poorly soluble and for these, it is necessary to construct biorelevant dissolution tests to forecast performance after oral administration

Dressman APV 2013

Dressman

Linking the lab to the patient…….

Hypothesis: the closer the test

conditions are to the GI physiology, the better the chances of predicting oral drug performance

0

200

400

600

800

1000

1200

1400

0 12 24 36 48 60 72

Time (hour)

Co

nc (

ng

/mL

)

Simulation (fasted)

Simulation (fed)

Fasted

Fed

GI parameters that affect product performance include:

Volume, composition (pH, surfactants etc.) and hydrodynamics of the fluids available to effect drug release from the dosage form

Transit characteristics of the dosage form through the GI tract

Biorelevant media

Stomach: FaSSGF: simulates low surface tension

& pH in the fasted stomach

FeSSGF: simulates gastric conditions after a standard PK study breakfast

Small intestine: FaSSIF-V2: simulates basal bile

secretion and pH in upper SI

FeSSIF-V2: simulates postprandial bile secretion, lipolysis products, increased buffer capacity and osmolality

Vertzoni et al. EJPB 2005, Dressman et al. Pharm.Res. 1998. Jantratid et al. Pharm. Res. 2008 APV 2013

Dressman

Media Simulating Fasted and Fed State

Stomach

Dressman

Biorelevant vs. compendial media: What is the difference?

Buffer capacity

Osmolality

Bile salts

Phospholipids

Lipolysis products

→ biorelevant

media for dissolution tests of poorly soluble drugs.

Media Simulating the Fasted and Fed State

Small Intestine

Biorelevant media

As these media are quite time-consuming to prepare in the lab, many scientists prefer to use „instant“ powders, commercially available from biorelevant.com

e.g. „Study of a Standardized Taurocholate–

Lecithin Powder for Preparing the

Biorelevant Media FeSSIF and FaSSIF“

Dissolution Technologies 2010

Dressman APV 2013

Dressman

Biorelevant media can predict food effects

Case example IV-IV-R: Nelfinavir mesylate

pKa : 6 and 11 (highly pH dependent solubility)

Cs : 2.16mg/mL in SGFsp (pH1.2) : 0.003 mg/mL in SIFsp (pH 6.8)

Log P : 4.0

Papp : 3.4x10-6 cm/sec (Caco-2)

BCS Class 2 (low solubility and high permeability)

Dressman

Dissolution in biorelevant media can predict food

effects: Case example nelfinavir mesylate

Dissolution of nelfinavir Mesylate 250mg tablets in 250ml media: Rate and extent of dissolution is much higher in fed state media (negligible in standard media)

0

10

20

30

40

50

0 30 60 90

Time (min)

Dis

solv

ed %

: FaSSIF-V2 : FeSSIF-V2 : SIFsp

: FaSSIF-old : FeSSIF-old

APV 2013

Dressman

Biorelevant media can predict food effects

Case example: Nelfinavir mesylate

Mean plasma plo file s o f ne lfin avir 1250mg

(250mg table t x 5 )

0

1000

2000

3000

4000

5000

0 6 12 18 24

Time (hour)

Conc (ng/

mL)

Fasted/in vivo

Fed/in vivo

4.9-fold increase in AUC

4.4-fold increase in Cmax

Recommended to be

taken with food

BCS Class 2 (low solubility and high permeability)

Linking the lab to the patient……

Biorelevant dissolution tests:

Qualitative forecast of food effects and formulation trends

But, how can we arrive at a more quantitative prediction??

IV-IS-IV-R:

Quantitative forecast of food effects and formulation trends possible, since contributions of all steps affecting bioavailability can be addressed

Dressman APV 2013

Dressman

Linking the lab to the patient……

Establish dosing

conditions and PBPK

parameters

Couple dissolution

results with PBPK

model

- Biorelevant media

(fasted/fed, gastric/small intestinal

media)

- Biorelevant dissolution parameters

(apparatus, volume, hydrodynamics)

=>Quantitative prediction based on PK

profile simulation, comparing in vivo data

(profile, AUC, Cmax, Tmax)

PBPK Models are needed to do this!

=>Qualitative prediction based on the

extent in dissolution in biorelevant media,

comparing with in vivo data (AUC, rank

order)

Testing with

biorelevant

dissolution method

Predict in vivo

performance

No

Refine

Yes

Refine

Refine

The in silico part of IV-IS-IV Modelling

PBPK (Physiologically Based Pharmacokinetic) Models

Models try to reflect all key aspects of the human physiology

Commercially available:

→ GastroPlus®, PK-SIM®, STELLA, SIMCYP®

VE

NO

US

B

LO

OD

P

OO

L

AR

TE

RIA

L B

LO

OD

P

OO

L

PANCREAS

SPLEEN

PORTAL VEIN

LIVER

KIDNEY

LUNG

DOSEi.v.

p.o.

GALL BLADDER

STOMACH

WALL

SMALL INTESTINE

WALL

LARGE INTESTINE

WALL

TESTES

HEART

BRAIN

FAT

BONE

SKIN

MUSCLE

Dressman

Model structure in STELLA program

Dissolved drug

in Stomach

Solid drug in

Small intestine

Dissolution in

intestinal media

Absorption => Drug in

Plasma

Gastric emptying

of liquid

Gastric emptying

of solid

Distribution and elimination

(Vd, K10, K12, K21)

Dissolution in

gastric media

Time (hour)

Pla

sm

a

co

ncen

trati

on

Drug

water

(meal)

Dressman

Plasma profile simulation based on

the STELLA program

• Input parameters

In vitro data

Dissolution rate (the gastric and small intestinal media)

Solubility (the gastric and small intestinal media)

Standard GI and dosing parameters

Dose

Co-administered water and meal volume

Gastric emptying rate

Post-absorptive disposition parameters

Distribution and elimination based on compartment model

• Assumptions in the model

- Negligible absorption from the stomach

- Simultaneous solid and liquid emptying from stomach

- intestinal permeability restrictions?

- drug precipitation upon arrival in the small intestine?

Dressman APV 2013

Case example IV-IS-IV-R: effect of particle

size on aprepitant pharmacokinetics

Dressman

Case example 1: Aprepitant

pKa : 9.7 (basic)

Cs : 0.02 mg/mL in SGFsp (pH1.2), 0.0007 mg/mL in SIFsp (pH 6.8)

Log P : 4.8

Papp : 7.8 x 10-6 cm/sec (Caco-2)

BCS Class 2 (low solubility and high permeability)

0

200

400

600

800

1000

1200

1400

0 12 24 36 48 60 72

Time (hour)

Co

nc (

ng

/mL

)

Fasted Fed

0

200

400

600

800

1000

1200

1400

0 12 24 36 48 60 72

Time (hour)

Co

nc (

ng

/mL

)

Fasted Fed

Micronized formulation Nanoformulation

Y. Shono et al. EJPB 76:95-104 (2010)

Dressman

Dissolution of aprepitant 125mg

in biorelevant media

Micronized Nanoformulation

0

10

20

30

40

50

0 15 30 45 60

Time (minute)

Dis

so

lve

d %

: FaSSIF-V2

: FeSSIF-V2

: SIFsp

0

10

20

30

40

50

0 2 4 6 8 10

Time (min)

% d

isso

lve

d

FaSSIF-V2

FeSSIF-V2

SIFsp

APV 2013 Y. Shono et al. EJPB 76:95-104 (2010)

Dressman

Simulated profiles of aprepitant

in the fasted and fed state

0

200

400

600

800

1000

1200

1400

0 12 24 36 48 60 72

Simulation (fasted)

Simulation (fed)

Fasted

Fed

0

200

400

600

800

1000

1200

1400

0 12 24 36 48 60 72

Time (hour)

Co

nc (

ng

/mL

)

Simulation (fasted)

Simulation (fed)

Fasted

Fed

Micronized Nanoformulation

Y. Shono et al. EJPB 76:95-104 (2010)

Value added: prediction of dissolution rate (=> particle size)

needed to eliminate food effect APV 2013

Dressman APV 2013

Case example 2: modeling example for a

development compound

„Compound A“ characteristics

Poorly soluble, lipophilic drug (log P 3.5)

Aqueous solubility is pH dependent: S >1g/mL at pH < 2

But: solubility drops drastically to values < 0.5µg/mL at pH > 4

Bioavailability low in rats (17%), dogs (27%), monkeys (4%)

Known Pgp substrate in animals

Available data: food effect study of 2 IR formulations (250mg and 500mg)

in 10 healthy volunteers. Food effect was found to be approx. 2.5 fold [ratio

AUC (fed/fasted) and cmax (fed/fasted)]

Absolute bioavailability unknown, no i.v. formulation.

N

NH

OH

NH

O2S

N

SCF

3

APV 2013

2

Step 1: BCS Classification

Solubility

• D/S ratio << 250mL at pH 1.2

• D/S ratio >>250mL at pH 4.5

and 6.8

→ poorly soluble

Permeability

No human jejunal permeability data

or data from mass balance

studies available.

Caco-2 permeability: 0.82·10-6

cm/sec

→ poorly permeable

Not only dissolution but also permeability are rate determining

towards the absorption of Compound A

IV III

II I

P

erm

eab

ility

Solubility

APV 2013

Step 2: Media Selection

Compendial…

Media simulating the stomach

SGF pH 1.2

Media simulating the small

intestine

SIF pH 6.8

(Compendial media do not reflect

the in vivo dissolution behavior of

poorly soluble drugs -> suitable

for highly soluble drugs)

…vs. biorelevant media

Media simulating the stomach

FaSSGF and FeSSGF

Media simulating the small

intestine

FaSSIF and FeSSIF

(Solubility and dissolution of poorly

soluble drugs can be enhanced by

bile salt/phospholipid(lipolysis

product mixed micelles → better

reflect the in vivo dissolution

behavior of poorly soluble drugs)

Step 3: Dissolution Tests

No significant difference between SGF

and FaSSGF, dissolution is fast and

complete.

Dissolution in FeSSGF shows lag time

(→ milk).

Dissolution in SIF and FaSSIF v2

negligible.

→ Compound A is clearly expected to

show a food effect.

→ Dissolution of Compound A is

favored in the fasted stomach but in the

fed small intestine.

0

20

40

60

80

100

120

0 50 100 150 200 250%

Dru

g R

ele

ased

Time [min]

SGF

FaSSGF

FeSSGF

FaSSIF (old)

FaSSIF V2

FeSSIF (old)

FeSSIF V2

SIF

Plateau concentration

Dissolution rate

APV 2013

C. Wagner et al. EJPB 82:127–138 (2012)

Step 4: Modeling

PBPK (Physiologically Based Pharmacokinetic)

Models try to reflect (all) aspects of the human

physiology.

Commercially available:

Gastro® Plus, PK® SIM, STELLA®, WinNonlin®

VE

NO

US

B

LO

OD

P

OO

L

AR

TE

RIA

L B

LO

OD

P

OO

L

PANCREAS

SPLEEN

PORTAL VEIN

LIVER

KIDNEY

LUNG

DOSEi.v.

p.o.

GALL BLADDER

STOMACH

WALL

SMALL INTESTINE

WALL

LARGE INTESTINE

WALL

TESTES

HEART

BRAIN

FAT

BONE

SKIN

MUSCLE

Results

Dissolution tests in Compendial media cannot predict the food effect of Compound A to any extent.

Dissolution tests in Biorelevant media can predict the food effect of

Compound A qualitatively but not quantitatively.

APV 2013

Results

IV-IS-IVC: by combining dissolution results in biorelevant

media with PBPK modelling, the

plasma profiles of Compound A can

be simulated quantitatively for two

dose strengths in both fasted and

fed state.

0

20

40

60

80

100

120

140

0 10 20 30 40 50

Time [h]

Pla

sm

a C

on

cen

trati

on

[n

g/m

l]

500mg in vivo (fasted)

500mg in vivo (fed)

500mg simulated, ka approach

(FaSSGF/FaSSIF v2)

500mg simulated, ka approach

(FeSSGF/FeSSIF v2)

0

10

20

30

40

50

60

0 10 20 30 40 50

Pla

sm

a C

on

cen

trati

on

[n

g/m

l]

Time [h]

250mg in vivo (fasted)

250mg in vivo (fed)

250mg simulated, ka approach (FaSSGF/FaSSIF v2)

250mg simulated, ka approach (FeSSGF/FeSSIF v2)

250mg IR formulation

500mg IR formulation

APV 2013

C. Wagner et al. EJPB 82:127–138 (2012)

In Vivo – in Silico Correlations

Statistical analysis indicates

that the in silico simulated

curves are equivalent to the in

vivo observed profiles.

With the more trasitional Levy

Plot analysis, an excellent

IVIVC is also obtained.

0

0.2

0.4

0.6

0.8

1

1.2

0 0.2 0.4 0.6 0.8 1 1.2

"Fraction absorbed" in vivo

"Fra

cti

on

ab

so

rbed

" in

sil

ico

y = 1.0182x – 0.0265

r2 = 0.9987

Fasted, 250mg IR formulation

“Levy Plot“

APV 2013

C. Wagner et al. EJPB 82:127–138 (2012)

Dressman

Strategy

Combine results from biorelevant dissolution tests with physiologically based pharmacokinetic modeling to

create

„in vitro – in silico – in vivo“

relationships

APV 2013

Some recent and current studies with in

vitro-in silico-in vivo Correlations

Aprepitant micronized vs nanosized API

Celecoxib food effects

Cinnarizine nanodispersed formulations

Dantrolene precipitation of sodium salt

Diclofenac enteric coated tablets and pellets

Fenofibrate micronized vs nanosized API & lipid formulations

Furosemide food effects

„Compound A“ food effects

Nelfinavir food effects

Nifedipine lipid, tablet and extended release formulations

Dressman

The Vision for 2020

To build a priori models which integrate dosage form performance, drug pharmacokinetic characteristics and gastrointestinal physiology, and

To use these models to predict plasma profiles in volunteers and patients accurately

=> better medicines, less risk to the patient and more streamlined development programmes in the pharmaceutical companies

APV 2013

Dressman APV 2013

Acknowledgments

Prof. Dr. Christos Reppas & Dr. Maria Vertzoni

(U-Athens)

Yasushi Shono (U-Frankfurt, Takeda)

Filippos Kesisoglou & Henry Wu (Merck & Co., Inc.)

…….and Thank YOU for your attention!