distribution gbm agile platform trial for - not ... agile... · gbm agile platform trial brian...

GBM AGILE Platform Trial

Brian Alexander, MD, MPHNEWDIGS DESIGN LAB: LEAPS PROJECT

July 18, 2018

CONFIDENTIAL -

not fo

r dist

ributi

on

Glioblastoma

• Incurable brain tumor– Radiation and temozolomide “standard”

• Most patients would like to receive an experimental therapy– Only ~10% enroll on trials Vanderbeek et al. 2018

• Despite much research, minimal advances over decadesCONFID

ENTIAL - no

t for d

istrib

ution

DRUG B

DRUG C

DRUG D

CONTROL DRUG A

DRUG E

DRUG B

Clinical practiceClinical research

CONFIDENTIAL -

not fo

r dist

ributi

on

Vanderbeek et al. Neuro Oncol 2018

CONFIDENTIAL -

not fo

r dist

ributi

on

www.lonza.com

Canadian Institutes of Health Research

CONFIDENTIAL -

not fo

r dist

ributi

on

Target selection and lead

validationPreclinical Phase I Phase II Phase III

IND NDACONFID

ENTIAL - no

t for d

istrib

ution

Target selection and lead

validationPreclinical Phase I

IND

NDA

Phase II Phase III

Phase II Phase III

Phase II Phase III

Phase II Phase III

Phase II Phase III

CONFIDENTIAL -

not fo

r dist

ributi

on

Target selection and lead

validationPreclinical Phase I

IND

NDA

Phase II Phase III

Phase II Phase III

Phase II Phase III

Phase II Phase III

Phase II Phase III

CONFIDENTIAL -

not fo

r dist

ributi

on

Biomarkers are different

Pao and Hutchinson. Nature Med 2012 TCGA Nature 2008

CONFIDENTIAL -

not fo

r dist

ributi

on

Targets are different

Vivanco et al. Cancer Discov 2012

CONFIDENTIAL -

not fo

r dist

ributi

on

Endpoints are different

Pseudo-response Pseudo-progression

Wen et al. J Clin Oncol 2010

Brandes et al. J Clin Oncol 2008Gilbert et al. NEJM 2014

CONFIDENTIAL -

not fo

r dist

ributi

on

Endpoints are different

Blumenthal et al. J Clin Oncol 2015 Han et al. Neuro Oncol 2014CONFID

ENTIAL - no

t for d

istrib

ution

Vanderbeek et al. Neuro Oncol 2018

CONFIDENTIAL -

not fo

r dist

ributi

on

CONFIDENTIAL -

not fo

r dist

ributi

on

Platform trial for GBM• Create one trial infrastructure

– Common biomarker, endpoint evaluations– Add/drop arms as trial is ongoing– Preserves indication-specific knowledge

• Replace serial trials– Eliminate “downtime”– Avoid “recreating the wheel”

• Ongoing concern to provide incentives for proactive therapeutic identificationCONFID

ENTIAL - no

t for d

istrib

ution

GBM Development is different than other cancers

u Imaging-based endpoints can be misleading

u Historical control comparison can under- and overestimatetreatment effect leading to bad decisions

u Different targets, different biomarkers for similar pathways

u Biomarker subgroups are not well defined• Overlapping genomic groups

• No gold standard for MGMT

Use OS

Use controls

GBM-specific agents

Multiplex markers, learn in trial

CONFIDENTIAL -

not fo

r dist

ributi

on

CONFIDENTIAL -

not fo

r dist

ributi

on

Subtypes with Stratification Biomarkers Only

Subtypes with Enrichment Biomarkers

Subtype• Characteristic of the patient/tumor• Partitions the biomarker space• Where randomization probabilities can

vary

Newly diagnosed methylated

Newly diagnosed

unmethylated

Recurrent disease

NDM+

NDU+

RD+

RD-NDU-

NDM-

CONFIDENTIAL -

not fo

r dist

ributi

on

Subtypes with Stratification Biomarkers Only

Subtypes with Enrichment Biomarkers

Newly diagnosed methylated

Newly diagnosed

unmethylated

Recurrent disease

NDM+

NDU+

RD+

RD-NDU-

NDM-

Signature• Characteristic of the experimental therapy• Can be overlapping• Best indication for further development

CONFIDENTIAL -

not fo

r dist

ributi

on

Subtypes with Stratification Biomarkers Only

Subtypes with Enrichment Biomarkers

Newly diagnosed methylated

Newly diagnosed

unmethylated

Recurrent disease

NDM+

NDU+

RD+

RD-NDU-

NDM-

Signature• Characteristic of the experimental therapy• Can be overlapping• Best indication for further development

CONFIDENTIAL -

not fo

r dist

ributi

on

Subtypes with Stratification Biomarkers Only

Subtypes with Enrichment Biomarkers

Newly diagnosed methylated

Newly diagnosed

unmethylated

Recurrent disease

NDM+

NDU+

RD+

RD-NDU-

NDM-

Signature• Characteristic of the experimental therapy• Can be overlapping• Best indication for further development

CONFIDENTIAL -

not fo

r dist

ributi

on

Subtypes with Stratification Biomarkers Only

Subtypes with Enrichment Biomarkers

Newly diagnosed methylated

Newly diagnosed

unmethylated

Recurrent disease

NDM+

NDU+

RD+

RD-NDU-

NDM-

Signature• Characteristic of the experimental therapy• Can be overlapping• Best indication for further development

CONFIDENTIAL -

not fo

r dist

ributi

on

Advantages of GBM AGILE for Drug Development

uGBM AGILE platform• Regulatory buy-in• Access to diverse group of experts in the field, operational expertise

enabling trial to evolve with success• Common infrastructure: harmonization of imaging, tissue acquisition,

trial data

uCost - commercial grade development vehicle at reduced Phase II/III pricing• Efficiencies in patient utilizations: common controls, adaptive

randomization• Right-sized” (# of patients = outcome of utility)

CONFIDENTIAL -

not fo

r dist

ributi

on

Advantages of GBM AGILE for Drug Development

u Speed• Accelerates time from “hand shake” to arm initiation to last patient out

• “Seamless” transition to Phase III, Phase II patients are used in final analysis

u Defines subsets of patients most likely to benefit from therapy

u Can incorporate predictive biomarkers (hypothesis testing and generating)

u *Forms the foundation for future combinations through the time machine/non-concurrent controls

CONFIDENTIAL -

not fo

r dist

ributi

on

CONFIDENTIAL -

not fo

r dist

ributi

on

DRUG B

DRUG C

DRUG D

CONTROL DRUG A

DRUG E

DRUG B

Clinical practiceClinical research

CONFIDENTIAL -

not fo

r dist

ributi

on

Clinical practice

Clinical research

Adaptive platform

trials

CONTROL DRUG A DRUG B DRUG C DRUG D DRUG E DRUG E

“Real world evidence”

CONFIDENTIAL -

not fo

r dist

ributi

on

Learning system

CONFIDENTIAL -

not fo

r dist

ributi

on

Doctors

Patients

Sponsors Regulators

Payers

CONFIDENTIAL -

not fo

r dist

ributi

on

THANK YOU

CONFIDENTIAL -

not fo

r dist

ributi

on

CONFIDENTIAL -

not fo

r dist

ributi

on