Slide 1

Session 5: Niche brokering: finding new users for old knowledge

How Personalized Healthcare May Bring New Life to Failed Drugs

OECD Expert Workshop on Knowledge Markets in Life SciencesNational Academies of Science

Washington, D.C.16-17 October 2008

2

Personalized Healthcare

Generating meaningful segmentation of patient populations, by whatever technology is appropriate (genomic, imaging, informatic), in order to increase the benefit of therapy

3

Everybody responds to therapy differently

Who suffers when therapies don’t work?

Patients

Physicians

Payers

Percentage Non-responders

4

Targeted Clinical Trials

Estimated that drug companies could save up to $100’sM per drug by incorporating pharmacogenomics data into trials

Trials terminated early due to success in a genetically defined subset of breast cancer patients Interim analysis yielded

statistically significant results, exceeding the primary endpoint

5

Access to the right therapy

Severe

Symptoms

Moderate

Symptoms

Mild Symptoms

Patient Population= Predicted Responders

= Predicted Low Efficacy

or Side Effects

6

Quicker uptake of therapeutic value

Arimidex vs. Herceptin: Global Sales from launch

0

500

1,000

1,500

2,000

2,500

3,000

3,500

Year

1

Year

2

Year

3

Year

4

Year

5

Year

6

Year

7

Year

8

Year

9

Year

10

Year

11

Year

12

$m

Traditional

PHC

TASCO presentation of

adjuvant data (May 05)

Adjuvant approval

EMEA May 06, US

FDA Nov 06

Initial adjuvant

indication in US

5 year analysis of

ATAC data

Breast Cancer Therapies: Global Sales from Launch

7

Towards Preventive Medicine

Avoiding futile medicine

Predictable therapeutic response

Earlier intervention

Delay onset and minimize severity

8

Rescue Strategies

The Good: Herceptin/Herceptest

The Bad: Herceptin/Herceptest

The Ugly: Herceptin/Herceptest

The Really Ugly: Iressa

9

The Good - Herceptin

Large responder effect in pivotal trials

FDA will not approve without Dx

Genentech develops IHC (Immunohistochemistry) with Dako

Receives approval and launches Herceptin/Herceptest

First blockbuster Personalized Therapy

Arimidex vs. Herceptin: Global Sales from launch

0

500

1,000

1,500

2,000

2,500

3,000

3,500

Year

1

Year

2

Year

3

Year

4

Year

5

Year

6

Year

7

Year

8

Year

9

Year

10

Year

11

Year

12

$m

Traditional

PHC

T

ASCO presentation of

adjuvant data (May 05)

Adjuvant approval

EMEA May 06, US

FDA Nov 06

Initial adjuvant

indication in US

5 year analysis of

ATAC data

Breast Cancer Therapies: Global Sales from Launch

10

The Bad - Herceptin

Genentech develops second test based on FISH (fluorescence in situ hybridization) with Vysis (now Abbott Molecular)

But: According to guidelines released by ASCO and the College of American

Pathologists, analyses of prospective, randomized, adjuvant trials of Herceptin show that testing algorithms to gauge HER2 expression have not been standardized and were developed “somewhat arbitrarily.”

An expert panel found “as many as 15 percent to 20 percent of the HER2 assays performed in the field may be incorrect when the same specimen was reevaluated in a high-volume, central laboratory.”

“Current testing methods for determination of the likelihood of benefit from Herceptin are not adequate,"

Third test, CISH (Chromogenic In Situ Hybridization), now available from Invitrogen

New test in the works from Monogram based on a proximity-based assay

11

The Ugly - Herceptin

Herceptin and HER-2 tests are reimbursed separately

Use of the HER-2 tests is not required by the label

Poor reliability and reduced trust in the HER-2 tests results in poor use of the tests by oncologists

Adverse (ultimately overturned) cost-effectiveness decision by NICE on use of Herceptin without HER-2 testing

12

The Really Ugly - Iressa

Iressa shows high levels of efficacy against non-small cell lung cancer in early trials

Very clear responder population in pivotal trials results in very limited label

AstraZeneca retracts the drug for all but compassionate use

Two academic labs suggest a role for EGFR in predicting drug response for Iressa and Tarceva

Iressa becomes the poster child for linked RxDx

But biological complexity rears its ugly head… Subsequent research indicates EGFR positivity doesn’t predict for

drug response, that the biology is much more complicated

Still no predictive test for Iressa

13

The Lesson – Herceptin/Iressa

Assuming the biology works:

No standards for sensitivity/specificity of Dx tests linked to therapeutics

Coverage/reimbursement decisions by public payers for new generation Dx, much less RxDx, antiquated and don’t provide ROI for clinical utility

Private payers want demonstration of clinical effectiveness

FDA indicating it will regulate this space

Business model for new generation Dx may be untenable

Could the Herceptin story replay today? Possibly

Could the Herceptin story replay tomorrow? Probably not

Retrofits of a Dx to a developed Rx are hard, start early

14

The PHC Landscape Today

PM has broad support but

will require changes in

how personalized

technologies are

evaluated, how health care

is financed and delivered,

and how clinicians and

consumers are prepared

Policy-makers,

researchers and

developers of PM

interventions need to

develop strategies to

overcome barriers to

clinical integration

Multi-stakeholder

collaborations, including

public-private

partnerships, will be

required

Stakeholders require

extensive and on-going

education in PM, in

particularly the

payer/purchaser

community and health

care delivery

organizations/providers

Source: Duke/ABT/PMC Landscape Analysis

15

Future Issues for PHC

Intelligent

legislation/

regulation

Viable

economic/

business

models for PM

offerings

Evidence

requirements

and reformed

reimbursement

policies

reflecting PM

product value

PM-savvy

health care

workforce

Enabling

technologies,

such as HIT

Source: PMC Strategic Plan

16

Repurposing Old Knowledge Informatic PHC

Can a large and fully integrated Electronic Health Record System (EHR) be used to demonstrate the value of antidiabetic therapy, in terms of comparative benefit and risk, in an environment reflecting actual clinical use of the therapy?

17

Cohort

>18 years old

Diagnosis of Diabetes in EHR.

Prescribed single oral hypoglycemic Agent.

– Patients with the metabolic syndrome and polycystic

ovarian syndrome excluded

– Patients on AGI excluded (n=146)

Baseline = Earliest Date of Hypoglycemic RX in

a diabetic patient.

Pts without follow-up were censored for all

outcomes except mortality.

18

Predictors

Measurements: Lipids, BMI, BP, LVEF, Hba1c,

eGFR, LFTs, Alb:Cr

Medical History: CHF, Liver Disease, Kidney

Disease, CAD, Stroke/TIA, Hepatitis B/C, Atrial

Fibrillation, New Diabetic?

Demographics: Age, Gender, Race

Oral Hypoglycemic Medication Class:

Meds: ACE / ARB, Cholesterol, ASA, Plavix

19

Medication Classes

Abbreviation Class Examples

“Big” biguanidesGlucophage (metformin)

“TZD” thiazolidinedionesAvandia (rosiglitazone)

“SFU” sulfonylureasGlucotrol (glipizide)

Dphen & MegD-phenylalanine derivatives. AKA. Meglitinides

Starlix (nateglinide)

“AGI”Alpha-glucosidase

Inhibitors

Precose (acarbose)

20

Outcomes

Nephropathy

Renal Insufficiency

Mortality

Coronary Artery Disease

Stroke/TIA

Liver Injury

Heart Failure

Continuous Outcomes: BMI, Hba1c, LDL, HDL, TG.

Anil Jain, MD Copyright 2007, Cleveland Clinic

Clinical Data

Repository

CCF DM

Data Set

Creating the Data Set…

CCF DM

Definitions

Social Security

Death Index

CCF has utilized an EHR since 2000.

EHR Tethered PHR ePrescribing Large Clinical Data

Repository 3.5 million patients 5 million prescriptions 89 million laboratory

results

Scope of Data Medications Demographics Diagnoses Procedures Laboratory Imaging

22

Conclusions

Successful use of the EHR to predict outcomes in type II diabetes

Models are very good at predicting clinical outcomes

Models for predicting continuous outcomes are better than simply relying on baseline alone.

Interactions allow for tailored treatment beyond AHRQ recommendations.

Potential to extend these techniques to other diseases.

23

24

Repurposing Old Knowledge for:Better understanding of the structure and interactions in

Clinical and Translational Science…STEM-01: Genetic medicine

STEM-02: Female

reproductive cell

biology STEM-03: Vascular disease

& control of cell proliferation

STEM-04: Controls

in early developmentSTEM-05: Gene

and cell therapy

STEM-06: Stem

cell therapy

STEM-07: Control

of hematopoiesis

STEM-08: Cell transplantation

& Tissue regeneration

STEM-09: Genetic

control of cell

differentiation

STEM-10: Diabetic

cell therapy

STEM-11: Molecular biology

of cell differentiation

STEM-12: Signaling

pathways in cell differentiation

STEM-13: Early development

and organ differentiation

STEM-14: Neural

regeneration

Stem Cell Science

25

Repurposing Old Knowledge for:Distinctive Competencies for Clinical and Translational Science

-omic understanding of disease

Slide 26

Thank-you

Wayne A. Rosenkrans, Jr., Ph.D.

Distinguished Fellow, MIT Center for Biomedical Innovation

Chairman & President, Personalized Medicine Coalition

Chief Applications Officer, SciTech Strategies

rosenkran1@aol.com

rosenkra@mit.edu

rosenkrans@mapofscience.com