Dissolution coupled with oral absorption modeling to predict clinically relevant performance David Sperry, Nikolay Zaborenko and Kaoutar Abbou Oucherif Eli Lilly Third FDA/PQRI Conference on Advancing Product Quality March 22-24, 2017

© 2017 Eli Lilly and Company

Outline

Biorelevant tools useful for drug development

Integrating in vitro and in silico models to predict in vivo performance

Three case studies

22 March 2017 © 2017 Eli Lilly and Company 2

Approach to biorelevant methods

22 March 2017 © 2017 Eli Lilly and Company 3

In Vitro Apps • Biorelevant fluids • Phenomenological

measurements

Models • Mechanistic

description

Mechanisms • Parameterize

models • Help translate in

vitro → in vivo

Biorelevant product performance predictions

Best predictions

In Vivo Absorption Model

Biorelevant tools • In vitro apparatuses

– USP-type paddle methods – One-step and two-step – Artificial Stomach Duodenum (ASD) model – Scaled-down two-step dissolution – Microcentrifuge test Biorelevant fluids – SGF (FaSSGF, FeSSGF) – FaSSIF, FeSSIF – Simulated saliva

22 March 2017 © 2017 Eli Lilly and Company 4

See, for example: E. Jantratid and J. Dressman, “Biorelevant Dissolution Media Simulating the Proximal Human Gastrointestinal Tract: An Update,” Diss. Tech., (2009) 16(3), 21.

Two-Step Dissolution • 300 mL 0.01 N HCl • 75 rpm paddle

→ 20 min • 300 mL FaSSIF • 75 rpm paddle

→ 260 min

USP-type methods: Example

22 March 2017 © 2017 Eli Lilly and Company 5

One-Step Dissolution • 900 mL FaSSIF • 75 rpm paddle

ASD model

22 March 2017 © 2017 Eli Lilly and Company 6

Stomach Secretion Rate = 2 mL/min

Duodenum Secretion Rate = 2 mL/min

Resting Volume = 50 mL Dosing Volume = 200 mL

Stomach Emptying: First order decay (w.r.t. Volume) Emptying Half-life = 15 min

Duodenum: Constant volume = 30 mL

Stomach

Duodenum

ASD measurements

22 March 2017 © 2017 Eli Lilly and Company 7

Stomach

Duodenum

Spec.

Light

Spec.

Light

Drug Dissolution Dilution and Transfer Out

Duodenum Data – Key Output Upper intestinal drug concentrations can be used as a surrogate for drug Absorption.

Stomach Data – Context Additional information can be learned from stomach profiles.

Biorelevant tools

• Mechanisms – Rotating disk dissolution (RDD, intrinsic

dissolution) – Flow-through imaging – NMR – Focused beam reflectance measurement

(FBRM)

22 March 2017 © 2017 Eli Lilly and Company 8

Diffusion coefficient

• Intrinsic dissolution rate: Solve Navier-Stokes for a rotating disk (Levich equation)

• Stokes-Einstein: laminar viscous drag on a spherical molecule

22 March 2017 © 2017 Eli Lilly and Company 9

Diffusion coefficient by other methods

22 March 2017 © 2017 Eli Lilly and Company 10

♦ Flow-through UV Imaging (SDi300)

http://www.sirius-analytical.com/system/files/Sirius%20Application%20 Note%20304-13%20UK%20web.pdf

♦ Diffusion-Ordered Spectroscopy 2D NMR

Lasentec® FBRM® experiments

22 March 2017 © 2017 Eli Lilly and Company 11

half way btw wall and shaft

Front view Side view

~20o angle

~ 1 cm from top of paddle

High pH

Low pH

Intermediate pH

Slide courtesy of Carrie Coutant

Biorelevant tools

• Dissolution Models

22 March 2017 © 2017 Eli Lilly and Company 12

Granules Tablet Powder Disintegration Disintegration

Very Limited Dissolution

Limited Dissolution

Best Dissolution

Limited physical models -> simple 1st-order rate

Limited physical models -> simple 1st-order rate

Good physical models

Hydrodynamics

Dissolution model • Noyes-Whitney dissolution model,

with several enhancements.

22 March 2017 © 2017 Eli Lilly and Company 13

0 10 20 300

0.5

1

t

Dissolved Coated tablet

Tablet

Granules

Aggregates

API

32134 rrrV π=

𝜕𝑟1,2,3

𝜕𝜕�min 𝑟1,2,3 >0

= −𝑘𝑡𝑡𝑡𝑡𝑡𝑡

𝜕𝑟𝑖𝜕𝜕�𝑟𝑖>0

= −𝑘𝑔𝑟𝑡𝑔

Tablet and granule disintegration

Henderson-Hasselbalch, pH and buffer capacity

API↔API+

HPO42-

H2PO4-

PO43- pH

Sol

parti

cle

Surface concentration

0.1 1 10 1000

0.05

0.1Particle size distribution

mi

r0i

Population balance

In vitro-in silico approach

22 March 2017 © 2017 Eli Lilly and Company 14

Biorelevant product performance predictions In Vitro Apps

Models Mechanisms

In Vivo Absorption Model

Johnson model

Input CR profile z-factor Fitted PSD Fitted D Mechanistic

parameters

Case 1 – Modified release formulation • Overall development challenge

– Increase the throughput of the process by a simple formulation change: increase the drug-containing layer thickness thereby increasing the drug load.

22 March 2017 © 2017 Eli Lilly and Company 15

• Formulation A – Spray-coated bead – Bead enteric coated

• “New” high-drug load formulation B – Drug load increased ~50% by spray-

coating more drug on bead – Other aspects remain the same

Core

Drug layer

Enteric coat * Some details of formulation omitted for clarity

Sperry et al., Molecular Pharmaceutics, 2010, 7, 1450–1457.

In vitro dissolution • Different strengths tested in vitro by different

methods

• Many method conditions explored – e.g. Both paddles and baskets specifically

investigated.

• Dissolution is different between formulations • And difference is not an artifact of the test

22 March 2017 © 2017 Eli Lilly and Company 16

A

B

Sperry et al., Molecular Pharmaceutics, 2010, 7, 1450–1457.

In vitro modeling

• Agreement suggests differences in release rate observed in vitro are due purely to the difference in surface area.

22 March 2017 © 2017 Eli Lilly and Company 17

Sperry et al., Molecular Pharmaceutics, 2010, 7, 1450–1457.

( ) ( )

−⋅

+−⋅⋅

−⋅⋅⋅⋅−

=VXCrrr

XX

rrhXD

dtdX d

sccs

c

s3

2

3330

033

0

03ρ

0.1 1 10 1000

0.05

0.1Particle size distribution

Mas

s in

bin

mi

r0i

rc

r0

Bioequivalence study • Met bioequivalence criteria

of 0.8 and 1.25

In this case, simple buffers and USP II method produced adequate biorelevant predictions.

22 March 2017 © 2017 Eli Lilly and Company 18

Sperry et al., Molecular Pharmaceutics, 2010, 7, 1450–1457.

Case 2 – Free base conversion • Solid forms: free base & a salt • Properties: Low solubility, pka ~7

22 March 2017 © 2017 Eli Lilly and Company 19

FaSSIF Salt Dissolution

Conversion to free base

Rotating Disk Dissolution

Free base

Salt

ASD: Salt supersaturation and precipitation

Stomach concentrations Duodenum concentrations

22 March 2017 © 2017 Eli Lilly and Company 20

Low dose

High dose D0=450

D0=120

D0=0.1

D0=0.03

Distribution of gastric conditions

22 March 2017 © 2017 Eli Lilly and Company 21

Normal

With PPI population

Map of Gastric conditions

In Vivo Absorption Model

Dissolution input

Systems-based Pharmaceutics

22 March 2017 © 2017 Eli Lilly and Company 22

- Linking manufacturing excellence with patient performance

Alliance

Case 3 – Integrating ASD data • Low solubility base

22 March 2017 © 2017 Eli Lilly and Company 23

ASD Duodenum Concentration

In Vivo Absorption

Model

In vitro ASD

In silico ASD

Dissolution/precipitation parameters

Precipitation rate parameters

22 March 2017 24

Model parameter Final value Initial guess 95% C.I Standard deviation

Growth constant 0.55 0.36 1.42 0.70 Growth order 1.60 0.81 2.29 1.13 Nucleation coefficient 14.01 13.85 974.9 481.2 Nucleation order 2.61 2.77 399.3 197.1

© 2017 Eli Lilly and Company

Duodenal Concentration Profile

Incorporating ASD results in gCOAS GI model

• 15 mg dose – Precipitation is predicted to be negligble. Fa =1

• 50 mg dose – Fa =0.82 due to precipitation in the

small intestine 22 March 2017 25

50 mg dose

15 mg dose

© 2017 Eli Lilly and Company

Acknowledgements

• Lee Burns • Carrie Coutant • Todd Gillespie • Brian Winger

22 March 2017 © 2017 Eli Lilly and Company 26