Roche Oncology Event, June 19, 2006

WorkshopRoche Diagnostics and Biomarker Development

Joachim EberleHead of R&D, Roche Centralized Diagnostics

2

Biomarkers and Roche

Making a Diagnostics test

Current Programs

Roche Oncology Event, June 19, 2006

3

Why is Roche interested in Biomarkers?

• Patients & Clinicians– early detection can increase chance of cure– patients can be selected to maximize

benefit and minimize toxicity• Payers

– optimized therapy allows more efficient use of straining healthcare budgets

– increased cost benefit per patient to reach “threshold”

• Regulatory– patient selection may be essential to gain

regulatory approval• Pipeline & Positioning

– allows us to make better development and portfolio decisions

5

Like

lihoo

d tr

eatm

ent

succ

ess

lower

higher

higherCost Effectiveness

treating patients that will benefit - not treating those that will not

Genom

ics

Prot

eom

ics

Future

4

Prognostic

Biological markers or “Biomarkers”Tests measuring a persons health status

Predict the likely response to a drug to discriminate “responders” from “non-responders”

Once disease state established, predict the probable course of the disease (“slow” vs. “fast” progressors)

Monitor efficacy of treatment or recurrence of disease

Discriminate “healthy” from asymptomatic “disease” state. Often used to screen large populations

Gene setup predisposes for disease - may or may not result in actual development of the disease

Risk Assessment / Predisposition

Therapy Monitoring

Screening / Early Detection

Patient Stratification/ Therapy selection

Roche Oncology Event, June 19, 2006

5

Risk Assessment

Screening / Early Detection

Normal - Pre-cancer - Early cancer - Carcinoma-Metastasis - Recurrence

Adjuvant Therapy Palliative Therapy

Personalized Treatment (Stratification)

Adaptationof Therapy

Potential use of Biomarkers in Oncology

Prognosis / Therapy Selection

Monitoring Recurrence, Therapy

Curative Intervention

Tum

or M

ass

(Cel

ls)

108

104

Cancer Progression

6

Biomarkers and Roche

Making a Diagnostics test

Current Programs

Roche Oncology Event, June 19, 2006

7

Roche Diagnostics R&D StrategyDrive growth through high-value products & services

• Development of existing platforms– markers – systems– services

time

sale

s

• Identification of new markersthrough initiatives in Genomicsand Proteomics

• Development of tests with differentiated clinical utility

Novel Products

8

Roche Oncology Biomarker Program For all clinical candidates, throughout their lifecycle

DiagnosticsDevelopment of prototype &commercial assaysAssessment new technologies

Exploration mode-of-action mechanismsBiomarker-guided development of new drugs

Predictive Biomarkers for drug selection/ differentiationSample sharing

PharmaBiology & clinical expertise

Health economicsCommercialization of drugs

Our competitive edge: sharing core expertise & development of targeted drugs

Roche Oncology Event, June 19, 2006

9

Herceptin

Tarceva

Xeloda

MabThera

Avastin

Pan-CDKInhibitor (R547)

Ocrelizumab(R1594)

MAI (R1530)

Epothilone D(R1492 & R1645)

Biomarker projects in support of Oncology projects

Research assays Validated assay In-vitro Diagnostic assay (if applicable)

Ph II Ph III / MarketPh I

- HER2 expression, gene copy numberand exploratory candidate markers

- EGFR gene copy number, expressionand mutations

- K-ras mutations and other exploratorymarkers

- Polymorphisms in in-activating enzymes

- FcgR-IIIa polymorphisms- ZAP-70 expression (CLL)

Omnitarg(R1273)

Range of candidate response markers

- Tissue collection for eventualinvestigation of future markers

Exploratory pharmacodynamic and

predictive response markers for all programs

10

GenomicsRNA/ DNA

ProteomicsPeptides / Proteins

Biomarker candidates• Technology (target) feasibility study• Tumor tissue, serum, urine, CSF etc. • Assay robustness evaluation• Technology implementation • IVD development

Roche Oncology Biomarker Program Two pathways to maximise marker detection and validation

Roche Oncology Event, June 19, 2006

11

Bladder

Lung

Liver

Blood / bone marrow

Prostate

Colon

Breast

MonitoringTherapy

predictionRelapse

predictionClassificationScreeningCancer site

Research program based on clinical need, medical value and technology match

P

#targets 1000 – 100,0001-10

G

MicroarraysTaqMan/ PCR Elecsys IMPACT (Protein Array)

P

1- 3 Up to 20

P

P

P G

G

G

G

Genomics Proteomics

PG P

G

P P

G

G

P

P G

G G

GG P P

G

12

From diagnostic marker candidate to a product Three development phases

• Internal R&D• External R&D • Co-operations• In-licensing • Markers in

public domain

Phase I:IdentifyingCandidates

2 Years

Marker Candidate100 candidates

Phase II:Prototype

Development

• Antibody dev.• Pre-validation• Prototype assays• Validation

studies on sample panels

1-1.5 Years

Validated Candidate20 candidates

• Test development• Platform • Manufacturing• Clinical trials • Registration• Marketing

Phase III:Product

Development

3 Years

Commercial Product~5 potential markers

Roche Oncology Event, June 19, 2006

13

Comparison of diseased and healthy tissue:

marker candidateDiscovery

1st filterWestern blots (tissue lysates)

Immuno-histochemistry

Raise antibodiesValidation

Tumor Control

Proteomics: From marker discovery to validated marker

2nd filterPrototype ELISA- marker in blood?

- sensitivity, specificity→ Panel A + Panel B

14

The composition of sample banks is the most critical factor during the validation process

Panel A:50 diseased patients compared to 50 healthy individuals

Panel B:1,600 samples (statistically relevant subgroup formation)

Age and gender matched All disease stages

Relevant disease controls according to prevalenceSpecificity controls (diseases not within the indication area)

Healthy controls

+-

discontinue validation

Roche Oncology Event, June 19, 2006

15

Our competitive advantage: a well-defined sample bank covering different stages

• Clinical samples have to be collected in intended indication before prototype assay development starts

• Serum and plasma samples collected. For specific indications other body fluids like urine, SF, CSF etc. are meaningful

• Clinical data are collected according to standardized CRFs

• A positive ethic committee approval and informed consents are mandatory

• Approx. 1,000 different samples are necessary for each indication due to statistical requirements

16

Biomarkers and Roche

Making a Diagnostics test

Current Programs

Roche Oncology Event, June 19, 2006

17

Overview: blood-based screening test programsDual development genomic and proteomic targets

> 90lung

no program> 20 candidate markersprostate

> 13discovery underwaybreast

> 12 22 candidate markerscolorectal

Protein markersMethylated DNA markersCancer

7 CRC & 2 Breast markers currently in validation

18

Colorectal cancer (CRC) screening Highly curable when diagnosed in early stage

Less than 40 % CRC detected at early stages

Urgent need for screening markers in body fluids indicating early stage disease

colon

small intestine

rectum

Location

Stage at first diagnosis:5-year survival:

Stage IV Distant

metastases

17 %5 %

Stage IIIRegional

lymph node involvement

28 %55 %

Stage I + IIRestricted

to gut

55 %90 %

Roche Oncology Event, June 19, 2006

19

Risk population ( > 50 yrs or symptoms orfamily history)

unsatisfactory as first-line screening assay

- highly invasive- poor acceptance- not useful as general

screening method

Sensitivity close to 100 %

Specificity 100 %

Sensitivity* 25-50 %(number of true CRC positives detected by the assay)

Specificity* 80-95 %(number of true negatives correctly identified)

Colonoscopy

FOBT(fecal occultblood test)

Confirmation by Biopsy

Current CRC screening algorithmsIssues with sensitivity, specificity &/ or compliance

* FOBT sensitivity: Collins et al. Annals Internal Medicine 2005; 142: 81 Imperiale et al. NEJM 2004; 351:2704

20

CRC Results: Final diagnostic test most likely will be a combination of markers

• Identified differences between "healthy" and diseased tissue

• Differences could be confirmed in serum using prototype ELISA

• Combinations of markers increases sensitivity

• Continue validation of additional markers

• Clinical validation underway

Receiver Operator Curves (ROC)of different markers:

AUC = area under the curve

Sens

itivi

ty %

0204060801000

20

40

60

80

100

Specificity %

CRC1 Marker(AUC= 0,801)[most significantsingle parameter]

INDEX CRC(AUC= 0,938)

Contribution by additional 13potential tumor markers

Roche Oncology Event, June 19, 2006

21

Morphology Cytogenetics Immunophenotyping

Cytochemistry FISH Molecular Biology

61-90 %

70-90 %

75-92 %

80-90 %

76-95 %

83-90 %

Current standards in diagnosis of leukemiaVarying reproducibility rates of gold standards

• No single test currently sufficient to establish diagnosis• Current tests subjective, with up to 7 days turnaround• Lack standardization & automation - need highly skilled staff

22

Genomics program: Leukemia microarrayImproved diagnosis enables better treatment

• Heterogeneous disease caused by genetic abnormalities

• Distinction between chronic, acute and sub-classifications essential for successful treatment

– ~20 subgroups

AmpliChip Leukemia• single assay for sub-classification of major leukemia classes• potential to replace other methods/ technologies• research program ’05/ ’06

del(17p)del(13q)

Others

t(9;22)

t(9;22)t(11q23)/ MLL

t(8;14)T-ALL

T-ALL corticalt(1;19)

t(12;21)Hyperdiploid

ALL Others

t(8;21)t(11q23)/ MLLt(15;17)inv(16)ComplexNormal + Others

ALLALL

nBMnBM

NLNL

MDSMDS

Others

CLLCLL

AMLAML

CMLCML

Roche Oncology Event, June 19, 2006

23

Gold standard panel of tests

Morphology Cytogenetics Immunophenotyping

FISHCytochemistry PCR

Microarray-based gene expression profile

n = 4000 patients

* The MILE study (Microarray Innovations in Leukemia”) is conducted in collaboration with the European Leukemia Network and US participants

MILE Study*Compare Clinical accuracy of microarray test with standard leukemia laboratory methods

24

Marker set distinguishes >20 subclasses All clinically relevant subclasses identified, accuracy ~96 %

Subclass

Data will be used to design custom microarray for application of gene expression profiling for diagnosis & sub classification of leukemia

CML16CLL15AML complex aberrant karyotype14AML with normal karyotype + other abnormalities13AML with t(11q23)/ MLL12

c-ALL/Pre-B-ALL with t(9;22)8AML with t(8;21)9AML with t(15;17)10AML with inv(16)/ t(16;16)11

ALL with hyperdiploid karyotype7ALL with t(1;19)6ALL with t(12;21)5

Pro-B-ALL with t(11q23)/ MLL2c-ALL/Pre-B-ALL with t(9;22)3

No call17

T-ALL4

mature B-ALL with t(8;14)1ClassNo.

ALL, T-lineage, matureALL, T-lineage, corticalALL, T-lineage, immature (Pro)ALL, T-lineage, immature (Pre)

ZAP-70 negativeZAP-70 positiveunmutated IgVHmutated IgVH

Roche Oncology Event, June 19, 2006

25

Summary

• Biomarkers are becoming a requirement in future healthcare

• Biomarker discovery is similar to drug development in timelines and success

• New assays most likely a combination of markers

• Roche has extensive joint program for all drugs throughout theirlifecycle

• Our strategy should provide Roche competitive edge in developinginnovative products and services

26

Appendix

Roche Oncology Event, June 19, 2006

27

MILE study inter-lab dataHigh inter- and intra-laboratory reproducibility

Data from the training phase (proficiency testing) using all genes represented on HG-U133 Plus 2.0 array

High correlation and concordance ensures high quality data

Reproducibility and concordanceInter-site comparison

28

2007 2008 2009

New Cancer marker launches

Bladder cancer monitoring test

Leukemia microarray (classification) Research use

AmpliChip p53 test (mutational analysis) Research use

Screening test for colorectal cancer

Cobas TaqMan

Methylation on Cobas TaqMan

Microarray

Breast relapse prediction test (p53)

Lymphoma (classification)Impact (Protein Array)

Screening test for prostate and breast cancer

Breast cancer therapy (p53)

Launch dates are estimates only; US launches may be later than indicated

2006