Garret Hampton | VP, Oncology Biomarker Development

Pharmaceuticals Division

Roche Pharma Day 2015

Molecular Information

Personalized healthcare

A cornerstone of the Genentech / Roche strategy

2

Diagnostics Pharma

60% of pipeline programs are being

developed with companion diagnostics

4 out of

9 BTDs

enabled by a Dx

that identified

patients most

likely to benefit

Molecule

Actemra (Systemic sclerosis)

Venetoclax (R/R CLL 17p)

Atezolizumab (NSCLC)

ACE 910 (Hemophilia)

Esbriet (IPF)

Lucentis (DR)

Atezolizumab (Bladder)

Alectinib (ALK+ NSCLC)

Gazyva (1L CLL)

Significant advances in cancer biology

Multiple molecular subsets of disease

Unknown

MET splice site MET

Amplification

KRAS NRAS

ROS1 Fusions

KIF5B-RET

EGFR

ALK Fusions

HER2 BRAF

PIK3CA AKT1

MAP2K1

PD-L1

Expression

3

4

Molecular Information in oncology

Combination of molecular and patient data will enable change in R&D and clinical practice

✓ ︎ ✗ ✓ ︎ ✗

Smarter, more

efficient R&D

Better

patient care Patients

matched to

clinical trials

Treatment plan

selected

Molecular understanding of cancer Patient outcomes

Database &

Analytics interface

Multiple capabilities required

Comprehensive tumor analysis and longitudinal assessment

mRNA-

expression

Cell signatures,

targets

RNA

sequencing

DNA-mutation

& CNVs

Ex: EGFR, BRAF

DNA

sequencing

Protein-

expression

PDL1, other CI

targets

Multiplex

IHC

Comprehensive tumor analysis Continuous monitoring over time

Cell free tumor

DNA

Blood DNA

sequencing

Imaging

Ex: ImmunoPet

Imaging

Ex: EGFR, BRAF

5

FMI R&D collaboration

Our partnership with FMI is key to informing R&D by leveraging molecular information

6

Molecular Information

from FMI database and

GNE/Roche clinical trials

(FM1 & FM1 heme)

mRNA-

expression

Cell signatures,

targets

RNA

sequencing

DNA-mutation

& CNVs

Ex: EGFR, BRAF

DNA

sequencing

Protein-

expression

PDL1, other CI

targets

Multiplex

IHC

Comprehensive tumor analysis Continuous monitoring over time

Cell free tumor

DNA

Blood DNA

sequencing

Imaging

Ex: ImmunoPet

Imaging

Ex: EGFR, BRAF

1 3

Development of a

blood-based

molecular assays

2

Immunotherapy panel

development (DNA & RNA)

2 2

Clinical trial data

Roche database FMI clinical database

50,000+ patient data in FMI database

1

Roche / FMI database queries

Example: PIK3CA mutations in multiple cancers

7

Implementation of FMI panels in development

Focus on high-value clinical samples

Clinical trial data

Roche database FMI clinical database

Current prioritization*:

1. Trials with post-progression biopsy (with archival baseline samples)

2. Phase 3 trial** n>300

3. Phase 2 trial* n>100

4. Phase 1b combos

** Positive trial, evidence of activity or potential identification of subset / high value for

informing our pipeline (i.e., importance of disease setting / indication / treatment)

50,000+ patient data in FMI database

1

8

FMI R&D collaboration

Immunotherapy R&D collaboration

9

2

Immunotherapy panel

development (DNA & RNA)

mRNA-

expression

Cell signatures,

targets

RNA

sequencing

DNA-mutation

& CNVs

Ex: EGFR, BRAF

DNA

sequencing

Protein-

expression

PDL1, other CI

targets

Multiplex

IHC

Comprehensive tumor analysis Continuous monitoring over time

Cell free tumor

DNA

Blood DNA

sequencing

Imaging

Ex: ImmunoPet

Imaging

Ex: EGFR, BRAF

2

Checkpoint inhibitors

Most effective in “inflamed” tumors

Inflamed Non-inflamed

TILs

PD-L1 expression

CD8+ T cells

Genomic instability

Pre-existing immunity

2

10

PD-L1 IHC: Staining for TCs and ICs

Assay sensitivity critical in detecting both cell types

11

Tumor cells

(TCs)

Tumor and immune cells

(TCs and ICs)

WCLC 2015

1IMvigor 210 ECC 2015, 2POPLAR ECC 2015

e.g. NSCLC e.g. bladder

Immune cells

(ICs)

2

Patient selection enriches for benefit

PD-L1 selected lung (TC & IC) and bladder cancer (IC)

Bladder cancer: Overall survival* Lung cancer: Survival hazard ratio*

* Monotherapy data

ECC 2015

Time (months)

100

80

60

40

20

0

5 6 7 8 9 10 11 1 2 3 4 0

– IC2/3

– IC0/1

+ Censored

12

Ove

rall

Su

rviv

al

Median OS 6.7 mo (95% CI, 5.7-8.0)

Median OS Not Reached (95% CI, 7.6-NE)

0.1 1

In favor of docetaxel

0.73

1.04

0.59

0.54

0.49

Hazard Ratio

In favor of atezolizumab

0.2 1 2

TC3 or IC3

TC2/3 or IC2/3

TC1/2/3 or IC1/2/3

TC0 and IC0

ITT N=287

2

12

Bladder cancer: Overall survival

Example: Atezolizumab phase 1 data

in urothelial bladder cancer patients

100

80

60

40

20

0

-20

-40

-60

-80

-100

0 21 42 63 84 105 126 147 168 189 210 231 252 273 294

Time on study (days)

Why do these patients progress?

Why do these patients respond? C

hange in s

um

of lo

ngest dia

mete

rs

(SLD

) fr

om

baselin

e (

%)

Time (months)

100

80

60

40

20

0

5 6 7 8 9 10 11 1 2 3 4 0

– IC2/3

– IC0/1

+ Censored

12

Ove

rall

Su

rviv

al

Median OS 6.7 mo (95% CI, 5.7-8.0)

Median OS Not Reached (95% CI, 7.6-NE)

Benefit is not the same for every patient

Some patients with high expression of PD-L1 do not benefit – why?

2

13

Many types of data will be needed to inform patient care including:

Biomarkers for cancer immunotherapy

Key platforms for discovery and development

mRNA-

expression

Cell signatures,

targets

RNA

sequencing

DNA-mutation

& CNVs

Ex: EGFR, BRAF

DNA

sequencing

Protein-

expression

PDL1, other CI

targets

IHC

Cell free tumor

DNA

Blood DNA

sequencing

Ex: EGFR, BRAF

Other Dx &

patient data

Ex: Imaging,

outcomes etc

EMRs

PDL-1 IHC and

multiplex IHC

Mutation burden

& neo-epitope

prediction

Immune cell

types and

signatures

1 2

14

Gene expression & combination hypotheses

Understanding the biology of immune cells in tumors enables combination hypotheses

−2 −1.6 −0.67 0.22 1.1 1.6 2

2 1 0 -1 -2

cbind(1:nc)

Indication

Me

lan

om

a

SC

C

NS

CL

C

RC

C

Ade

no

NS

CL

C

Bla

dde

r

TN

BC

HE

R2

+B

C

CR

C

HR

+B

C

Th17 Signature

Treg Signature

Th2 Signature

Teff Signature

B cell Signature

IB Signature

IFNg Signature

NK cell Signature

APC Signature

Myeloid Signature

1

15

Gene expression & combination hypotheses

Understanding the biology of immune cells in tumors enables combination hypotheses

16

Myeloid signature (macrophages) associated with

lack of response to atezolizumab in bladder cancer

−2 −1.6 −0.67 0.22 1.1 1.6 2

2 1 0 -1 -2

cbind(1:nc)

Indication

Me

lan

om

a

SC

C

NS

CL

C

RC

C

Ade

no

NS

CL

C

Bla

dde

r

TN

BC

HE

R2

+B

C

CR

C

HR

+B

C

Th17 Signature

Treg Signature

Th2 Signature

Teff Signature

B cell Signature

IB Signature

IFNg Signature

NK cell Signature

APC Signature

Myeloid Signature

Anti-CSF-1R

Anti-PD-L1

Tumor associated macrophages

Hypothesis: Anti-CSF-1R removes macrophages

which may enable atezolizumab activity

Myeloid signature:

IL1B, IL8, CCL2

P=0.01

PD = Progressive disease

SD = Stable disease

CR/PR = Complete/partial response

1

T cell infiltration

Cancer T cell recognition

anti-CEA/CD3 TCB

anti-CD20/CD3 TCB

anti-HER2/CD3 TCB

ImmTAC* (Immunocore)

anti-VEGF (Avastin)

anti-Ang2/VEGF (vanucizumab)

Chen and Mellman. Immunity 2013; CI=cancer immunotherapy

T cell killing

anti-PDL1 (atezolizumab)

anti-CSF-1R (emactuzumab)

anti-OX40

IDOi

IDOi* (Incyte)

CPI-444* (Corvus)

anti-TIGIT

IDO1/TDOi* (Curadev)

EGFRi (Tarceva)

ALKi (Alectinib)

BRAFi (Zelboraf)

MEKi (Cotellic)

anti-CD20 (Gazyva)

anti-HER2 (Herceptin; Kadcyla; Perjeta)

various chemotherapies

lenalidomide

rociletinib* (Clovis)

Antigen presentation

T-Vec oncolytic viruses* (Amgen)

INFa

anti-CD40

CMB305 vaccine* (Immune Design)

Priming & activation

anti-CEA-IL2v FP

anti-OX40

anti-CD27* (Celldex)

entinostat* (Syndax)

Antigen release

17

Unlocking full value of CI through combinations

Broadest industry portfolio in oncology

1

Clinical development Preclinical development

* Partnered or external

Established therapies

Schmid et al ECC 2015

Patients with a high tumor IFNg-associated gene signature derive OS

benefit from atezolizumab in NSCLC (POPLAR)

PD-L1 expression

IFN-g producing CD8+ T cells

Genomic instability

Pre-existing immunity INFLAMMED

Gene expression as a predictive biomarker

Presence of INFg-producing CD8 T-cells predicts benefit in NSCLC treated with atezolizumab

1

18

Many types of data will be needed to inform patient care including:

Biomarkers for cancer immunotherapy

Key platforms for discovery and development

mRNA-

expression

Cell signatures,

targets

RNA

sequencing

DNA-mutation

& CNVs

Ex: EGFR, BRAF

DNA

sequencing

Protein-

expression

PDL1, other CI

targets

IHC

Cell free tumor

DNA

Blood DNA

sequencing

Ex: EGFR, BRAF

Other Dx &

patient data

Ex: Imaging,

outcomes etc

EMRs

Mutation burden

& neo-epitope

prediction

Immune cell

types and

signatures

1 2

19

PFS

/ O

S

Time

High

Low

Mutation burden

# o

f m

uta

tio

ns

e.g. Lung tumors

Low

High

1

2

Tumor cell

1

Tumor cell

2

Mutation burden is clinically important

Tumors with higher numbers of mutation tend to be more sensitive to anti-PDL1 / anti-PD1

Hypothetical – partly based on: Rivzi et a l Mutational landscape determines sensitivity to PD-1 blockade in non-small cell lung cancer,

Science, 2015; Snyder et al Genetic basis for clinical response to CTLA-4 blockade in melanoma, NEJM 2014

2

20

Conclusions

21

Personalised

healthcare

Molecular

Information

Transformational

time

Diagnostics Pharma

R&D insights

✓ ︎ ✗