Computational Systems Pharmacology of Antibody-Drug Conjugates: A - - PowerPoint PPT Presentation

Computational Systems Pharmacology

• f Antibody-Drug Conjugates:

A Joint Academia-Industry Experience

Inez Lam

IMAG-AND Futures Meeting March 17, 2020

Mac Gabhann Lab at Johns Hopkins University

C

COMPUTATIONAL DESIGN

• f
• f THERAPEUTICS LAB

Diverse Biological Contexts for Multiscale Modeling

Cancer Vascular Diseases HIV Endometriosis

Diverse Research Contexts for Multiscale Modeling

Johns Hopkins-AstraZeneca Scholars Program: A Joint Academia-Industry Experience

In Industry

Lots of data Drug development expertise Real-world application

Ac Academia ia

Long-term projects Big picture questions Modeling expertise Clinical expertise

Applying Quantitative Systems Pharmacology at the intersection of academia and industry

Co Computational Mo Models ls & Si Simul ulations

• ns

Ex Experiments

Da Data Pr Prediction

• ns

Antibody-Drug Conjugates: The Best of Both Worlds?

targets cancer cells joins antibody and drug kills cancer cells (also known as the warhead) High Specificity High Cytotoxicity

+ = Selective killing of tumor cells

Antibody Small molecule drug Linker

Why model ADCs?

Need to optimize 3 different entities to determine collective properties of the ADC: – Mu Multiple design levers can be controlled – Ph Pharmacokinetics and pharmacodynamics of individual components and overall ADC – Th Therapeutic Index: balance between safety & toxicity at multiple scales

Can be applied at any stage of drug development process

Clinical Preclinical Discovery

Why Model Antibody-Drug Conjugates?

Drug to Antibody Ratio (DAR): average number

• f drug molecules

attached to the antibody

Antibody Warhead Linker

Gr Groups ps No Notabl ble M Mode dels Sel Selec ected ted Insi Insights hts

Un Univers rsity ty

• f
• f

Mi Michigan Preclinical, multiscale model of T-DM1 (Ciliers 2016) Model of payload distribution (Khera 2017) Agent-Based Model of T-DM1 (Menezes 2020) Co-administration of unconjugated Ab with ADC may help improve distribution of ADC in the tissue

Existing models have explored various aspects of ADCs

Gr Groups ps No Notabl ble M Mode dels Sel Selec ected ted Insi Insights hts

Un Univers rsity ty

• f
• f

Mi Michigan Preclinical, multiscale model of T-DM1 (Ciliers 2016) Model of payload distribution (Khera 2017) Agent-Based Model of T-DM1 (Menezes 2020) Co-administration of unconjugated Ab with ADC may help improve distribution of ADC in the tissue SU SUNY NY Bu Buffal alo PKPD Model of brentuximab-vedotin (Shah 2012) PKPD model of inotuzumab ozogamicin (Betts 2016) Tumor Disposition Model for T-DM1 (Singh 2016) Dual Cell-Level Systems PKPD Model (Singh 2019) Model suggested fractioned dosing regimen is superior to a conventional dosing regimen for acute lymphocytic leukemia (ALL)

Existing models have explored various aspects of ADCs

Gr Groups ps No Notabl ble M Mode dels Sel Selec ected ted Insi Insights hts

Un Univers rsity ty

• f
• f

Mi Michigan Preclinical, multiscale model of T-DM1 (Ciliers 2016) Model of payload distribution (Khera 2017) Agent-Based Model of T-DM1 (Menezes 2020) Co-administration of unconjugated Ab with ADC may help improve distribution of ADC in the tissue SU SUNY NY Bu Buffal alo PKPD Model of brentuximab-vedotin (Shah 2012) PKPD model of inotuzumab ozogamicin (Betts 2016) Tumor Disposition Model for T-DM1 (Singh 2016) Dual Cell-Level Systems PKPD Model (Singh 2019) Model suggested fractioned dosing regimen is superior to a conventional dosing regimen for acute lymphocytic leukemia (ALL) Indus Industry Novartis ADC Modeling Framework (Vasalou 2015) Genentech ADC PK Model (Sukumaran 2017) Found undesirable tumor properties that can impair ADC tissue homogeneity and explored ADC design scenarios to counteract them

Existing models have explored various aspects of ADCs

Our Systems Pharmacology Model of PBD ADCs

Goal: Build a multiscale, integrated computational model of AstraZeneca pyrrolobenzodiazepine (PBD PBD) ADCs using differential equations with “bench to bedside” translation Use model to improve prediction of therapeutic index (TI) and to understand: – Properties of the ADC to build the optimal therapy – Kinetics and mechanisms of ADC action in tumor – Off-target and bystander effects – Factors that have the biggest impact on safety and efficacy Framework can be adapted to other types of ADCs

In In vivo da data In In vitro da data Cl Clinical data Co Computational Models

Cellular & Intracellular Mechanisms of ADCs Tumor Properties and Systemic Distribution Virtual Patients with Specific Attributes

In In Vitro Model In In Vivo Model Cl Clinical Model Si Simul ulations

• ns

Modify ADC Design Characteristics Predict Bystander Effects and Tumor Growth/Inhibition Run Virtual Clinical Trials and Test Different Treatment Scenarios

Li Link to Therapeutic Index of ADCs

Cell Surface Nucleus

Win

intra trace cellu llula lar

+ Wnuc

nuclea ear

DN DNA Wnuc

nuclea ear

DN DNA

–

Ag Ag

+

Lysosome

AD ADC

– Ag

Ag AD ADC

– Ag

Ag AD ADC Wex

extracel ellul ular

x DAR Endosome

– Ag

Ag AD ADC

+

AD ADC Ag Ag Ag Ag

Cytosol

In In Vitro Model Schematic

Bi Bindi ding g & Un Unbi bindi ding Ce Cell Death In Internalization & & Recycling Tr Trafficking & Release of Warhead Fo Formation

• n of
• f

Ef Effector

• r Com
• mplex

Cell Surface Nucleus

Win

intra trace cellu llula lar

+ Wnuc

nuclea ear

DN DNA Wnuc

nuclea ear

DN DNA

–

Ag Ag

+

Lysosome

AD ADC

– Ag

Ag AD ADC

– Ag

Ag AD ADC Wex

extracel ellul ular

x DAR Endosome

– Ag

Ag AD ADC

+

AD ADC Ag Ag Ag Ag

Cytosol

In In Vitro Model Schematic

Ce Cell Death Us Use In Vitro Mode del to Ex Explor

• re Ke

Key ADC De Design Properties

Drug to Antibody Ratio (DAR): average number

• f drug molecules

attached to the antibody

As As DAR AR increases: Wa Warhead-DN DNA complex increases linearly Ex Extracellular Warhead increases linearly Ce Cell Population decreases exponentially

Ex Explor

• ring Ke

Key ADC Desi sign Prop

• perties:

s: Varying DAR R between 1 and 10

Bi Bigge ggest ga gain in cell killing g from fir first fe few warhead mole lecule les: Sug Sugges ests op

• ptimal DAR for thi

his syste system may y be betw tween 2-4

Li Linker Design Wa Warhead Potency Sens Sensitivity Ana nalysis

Us Usin ing the In Vit itro Model l to Explo lore Key AD ADC C Desig ign Pr Propertie ies

kki

kill =

= 103 1/ 1/nM nM/hr hr

10 104 10 105 10 106 10 107 10 102 10 10

Blood Rest of Body Intravenous Dosing Clearance

ADC Ab + W

In Vivo Model Connects Pharmacokinetics to In Vitro Model

In In vivo da data In In vitro da data Cl Clinical data Co Computational Models

Cellular & Intracellular Mechanisms of ADCs Tumor Properties and Systemic Distribution Virtual Patients with Specific Attributes

In In Vitro Model In In Vivo Model Cl Clinical Model Si Simul ulations

• ns

Modify ADC Design Characteristics Predict Bystander Effects and Tumor Growth/Inhibition Run Virtual Clinical Trials and Test Different Treatment Scenarios

Li Link to Therapeutic Index of ADCs

Diverse contexts enrich the research experience

Developing long-term partnerships with both academic and industry researchers enables di diversity

• f
• f expertise

se and resou sources for multiscale modeling Harnessing the ad advan antag ages of each research environment Understanding both research environments will str strea eamline ne futur future e col

• llabora
• rati

tions

• ns

Thanks for listening! Questions?

Johns Hopkins-AstraZeneca Scholars Program Special Thanks To:

Feilim Mac Gabhann Rosalin Arends Kathryn Ball Phin Chooi Thaïs Callieau Balakumar Vijayakrishnan Peter Tyrer Conor Barry Sarvenaz Sarabipour Christy Pickering Wangui Mbuguiro Adriana Gonzalez