A Light in the Darkness… Peter J. O’Dwyer Abramson Cancer Center University of Pennsylvania.

A Light in the Darkness…

Peter J. O’DwyerAbramson Cancer Center

University of Pennsylvania

Disclosures

• Research funding: GSK, Genentech, Arqule, Pfizer, Novartis, Celgene, Incyte, Methylgene.

• Consultant: Genentech, Celgene

Glimmers…• DNA aberrations – many described, few

“actionable”• Relationship to outcome not always clear• Early versus late in carcinogenesis• Detected in primary tumor or metastasis• Associations with outcome in models vs

patients

Studies targeting defined populationsPharmacodynamic and efficacy analysis of the BRAF inhibitor dabrafenib(GSK436) in combination with the MEK inhibitor trametinib (GSK212) inpatients with BRAFV600 mutant colorectal cancer (CRC). (Abstract 3507)R. B. Corcoran, G. S. Falchook, J. R. Infante, O. Hamid, W. A. Messersmith, A.Daud, E. L. Kwak, D. P. Ryan, R. Kurzrock, C. E. Atreya, J. Luan, P. Sun, M.Schaeffer, M. Motwani, M. R. Bleam, C. H. Moy, K. Patel, K. W. Orford, S. Kopetz,A. P. Venook

A randomized, placebo-controlled, phase I/II study of tivantinib (ARQ 197) incombination with cetuximab and irinotecan in patients (pts) with KRASwild-type (WT) metastatic colorectal cancer (CRC) who had receivedprevious front-line systemic therapy. (Abstract 3508) C. Eng, L. L. Hart, A.Severtsev, O. Gladkov, L. Mueller, M. V. Kopp, V. I. Vladimirov, R. M. Langdon Jr.,B. Kotiv, S. Barni, C. Hsu, E. Bolotin, R. Von Roemeling, B. E. Schwartz, J. C.Bendell

Issues for each study

• Importance of target

• Reliability of the assay

• Measurement of effect

• Conclusions

BRAFi dabrafenib plus MEKi trametinib in BRAF V600 mutant colorectal cancer

• BRAF mutation does confer adverse prognosis in the MSS population

• However BRAF inhibition alone in BRAF-mutant tumors was associated with limited efficacy

• Inhibition of downstream signaling at MEK rational, and supported by preclinical data. Samowitz W S et al. Cancer Res 2005;65:6063-6069

Detection method – analytical, primary versus metastasis

• Eligibility established locally• Central analysis of samples by next-generation

sequencing, using Ion Torrent platform• Principally derived from archival primary

tissues• For some markers this may be adequate, but

heterogeneity of tumors, especially between primary and metastasis, will confound interpretation of results.

Results: EfficacyWaterfall Plot for CRC Patients (n=36*)

Series1

–100

–80

–60

–40

–20

0

20

40

60

80

100

Max

imu

m P

erce

nt

Red

uct

ion

Fro

m B

asel

ine

Mea

sure

men

t Best Unconfirmed ResponseComplete response Progressive diseasePartial response Not evaluableStable disease

Unconfirmed responses (CR+PR): 4 (11%)• 2 confirmed responses (1 CR and 1 PR)

Unconfirmed minor responses: 8 (22%)

Note: Horizontal reference lines are at –30% and 20%.*Non-evaluable subjects who are still in the study and have not progressed with no post dose disease assessments are excluded from the analysis**Minor response defined: reduction of between 10 – 30%

Difficult Endpoints / Essential for Progress• Salient feature of this trial was commitment to PD

endpoints

Biomarker analysesReduction in pERK observed in all patientsPI3KCA mutations do not preclude clinical activityMSI and low EGFR may be associated with better outcomes

EGFR ≤ 100EGFR > 100

% Change in pERK

294 1614 1572 1573 1613 1577 Average

-100.0

-80.0

-60.0

-40.0

-20.0

0.0

20.0

40.0

60.0

80.0

100.0

Individual patients

0 3 6 9 12 15

1.0

0.8

0.6

0.4

0.2

0.0

Time from First Dose (Months)

Prop

ortio

n Al

ive

and

Prog

ress

ion

free

EGFR ≤ 100EGFR > 100

Number at risk

Incremental Advance in BRAF-mutated Tumors

-100

-75

-50

-25

0

25

50

75

100

%C

han

ge

Fro

m B

asel

ine

(Su

m o

f L

esio

n S

ize)

Kopetz et al, ASCO 2010 Corcoran et al, ASCO 2013

Vemurafenib alone Dabrafenib/Tremetinib

Series1

–100

–80

–60

–40

–20

0

20

40

60

80

100

Max

imum

Per

cent

Red

uctio

nFr

om B

asel

ine

Mea

sure

men

t

Complete response

Progressive disease

Partial response

Stable disease

Corcoran RB, Ebi H, Turke AB, et al. Cancer Discov. 2012

Next Steps in BRAF-mutant Tumors

Concomitant BRAF and PI3K/mTOR Blockade Is Required for Effective Treatment of BRAFV600E Colorectal Cancer. Coffee EM, Faber AC, Roper J, et al. Clinical Cancer Res 2013

Resistance to BRAF inhibition in BRAF-mutant colon cancer can be overcome withPI3K inhibition or demethylating agents. Mao M, Tian F, Mariadason JM, et al. Clin Cancer Res, 2013.

1. Combined BRAF and EGF receptor tyrosine kinase inhibition

2. Combined BRAF and PI3K pathway inhibition

BRAFi dabrafenib plus MEKi trametinib in BRAF V600 mutant colorectal cancer

• Target: BRAF mutation confers poor prognosis in MSS tumors

• The clinical outcomes here suggest an incremental gain.

• Good-sized population – as more people perform mutational analyses on colorectal cancer patients (is there any justification for omitting Braf from analysis now?)

• As with single agent BRAF-directed therapy, progression follows a period of disease control – identifying molecular events in the progressing tumor is a priority.

• Difficult to do the biopsy on progression unless there is a therapeutic option for the patient.

Studies targeting defined populationsPharmacodynamic and efficacy analysis of the BRAF inhibitor dabrafenib(GSK436) in combination with the MEK inhibitor trametinib (GSK212) inpatients with BRAFV600 mutant colorectal cancer (CRC). (Abstract 3507)R. B. Corcoran, G. S. Falchook, J. R. Infante, O. Hamid, W. A. Messersmith, A.Daud, E. L. Kwak, D. P. Ryan, R. Kurzrock, C. E. Atreya, J. Luan, P. Sun, M.Schaeffer, M. Motwani, M. R. Bleam, C. H. Moy, K. Patel, K. W. Orford, S. Kopetz,A. P. Venook

A randomized, placebo-controlled, phase I/II study of tivantinib (ARQ 197) incombination with cetuximab and irinotecan in patients (pts) with KRASwild-type (WT) metastatic colorectal cancer (CRC) who had receivedprevious front-line systemic therapy. (Abstract 3508) C. Eng, L. L. Hart, A.Severtsev, O. Gladkov, L. Mueller, M. V. Kopp, V. I. Vladimirov, R. M. Langdon Jr.,B. Kotiv, S. Barni, C. Hsu, E. Bolotin, R. Von Roemeling, B. E. Schwartz, J. C.Bendell

Tivantinib (ARQ 197) with cetuximab and irinotecan in KRAS WT colorectal cancer

• MET amplification in 10% of colon primaries, but 8/9 hepatic metastases (di Renzo, Comoglio, 1995)

• More recent analyses suggest that MET aberrations in tumor not common: one amplification and 4 mutations among 212 colorectal tumors (TCGA, Nature, 2012)

• MET inhibition not effective in patient-derived xenografts (Galimi F, Clin Cancer Res, 2011)

• Oncogenic addiction to MET not established in colorectal cancer• Twin rationale: c-met inhibition, improved EGFR pathway

targeting

Detection method – analytical, primary versus metastasis

• Retrospective analysis of c-met expression• Eligibility not based on MET expression• Immunohistochemistry-based• Validated across several tumor types• Principally derived from archival primary

tissues

Tivantinib (ARQ 197) with cetuximab and irinotecan in KRAS WT colorectal cancer

• Trial design– Randomized Phase II– Early termination unfortunate – underpowered:“The number of PFS events for analysis was decreased from 110 to 80, which

provides 70% power to detect a 50% improvement in median PFS at P = 0.10 (1-sided)”

• Demographics – Eligibility not based on MET expression– Good PS, well-balanced, placebo had more non-

responders to prior therapy

Results

• PFS: 8.3 vs 7.3 months, P = 0.38• Response: 45% versus 33%• OS: 19.3 vs 16.8 months, P = 0.25

Trend toward tivantinib/cetuximab/irinotecan

Results• PFS: 8.3 vs 7.3 months, P = 0.38• Response: 45% versus 33%• OS: 19.3 vs 16.8 months, P = 0.25

• Areas of potential interaction– Pharmacokinetic – toxicity similar– Pharmacodynamic

• MET-related• MET-independent (though tivantinib has high specificity)

Tivantinib (ARQ 197) with cetuximab and irinotecan in KRAS WT colorectal cancer

• Immunohistochemical analyses on 67 patients (59 primary, 8 metastases)

• Antibody validated and accepted• Results unexpected:

Pro

gre

ssio

n-F

ree

Su

rviv

al,

%

0 3 6 9 12 15 18

100

50

75

25

0

Tivantinib(n = 24)Placebo(n = 20)

0 3 6 9 12 15 2118

100

50

75

25

0

Pro

gre

ssio

n-F

ree S

urv

ival,

%

21

HR = 0.22 (95% CI, 0.06 - 0.80)Log-rank P = 0.01

HR = 0.74 (95% CI, 0.36 - 1.52)Log-rank P = 0.41

Tivantinib(n = 11)Placebo(n = 12)

PFS ORR: T = 54.2%; P = 30.0% ORR: T = 27.3%; P = 41.7%

MET-High MET-Low

MET as Survival Signal in Colon Cancer• May permit survival in hostile environment (Vogelstein B, et

al. Nature, 2013)• In melanoma, HGF secretion by stromal cells confers

resistance to vemurafenib (Straussman R, et al. Nature, 2013)

HGF-High

Pro

gre

ssio

n-F

ree S

urv

ival,

%

0

100

50

75

25

0

Tivantinib(n = 29)

Placebo(n = 28)

HR = 0.70 (95% CI, 0.37 - 1.32)Log-rank P = 0.27

PFS

4 8 12 16 20 24

HGF-Low

Pro

gre

ssio

n-F

ree S

urv

ival,

%

0

100

50

75

25

0

Tivantinib(n = 30)

Placebo(n = 28)

HR = 0.94 (95% CI, 0.50 - 1.80)Log-rank P = 0.86

4 8 12 16 20 24

ORR: T = 44.8%; P = 25.0% ORR: T = 46.7%; P = 39.3%

Tivantinib (ARQ 197) with cetuximab and irinotecan in KRAS WT colorectal cancer

• Is MET a valid target in this disease?– Requires definitive trial in patients. Metastatic disease focus

supported by early literature. – Preliminary expression relationships differ from those in other

tumors, where high MET expressors are susceptible– For these reasons a follow-up trial should probably be based upon

MET expression in metastases• How best to select patients for future trials – FISH, IHC, other?• Is tivantinib a MET inhibitor? Recent data support an effect

on microtubules in vitro (Katayama R, et al. Cancer Res, 2013)

• As authors conclude, additional PD in this and other tumor types will define future pathway.

Is there light here?• Dependability of preclinical models as a guide

– The proper in vivo model for colorectal cancer is the patient

• Patient selection – rigorous validated analyses, often will require sample of metastatic disease

• How can this be achieved with many targets, many drugs?

Obstacles to a Genomic Approach to Colorectal Cancer Treatment

PATIENTS

GENOMES

TARGETED DRUGSCost of

sequencing

Cost of Trials

Regulatory barriers

Drug availability

Obstacles to a Genomic Approach to Colorectal Cancer Treatment

PATIENTS

GENOMES

TARGETED DRUGS

1. Accessing fractions of total patient volume – numbers.2. Bringing a next-generation sequencing platform to broad population. 3. Cost of performing trials one by one.

Obstacles to a Genomic Approach to Colorectal Cancer Treatment

PATIENTS

GENOMES

TARGETED DRUGS

1. Publicly- or privately-funded screening studies.2. Uniform next-generation sequencing platform. 3. Permits multiple studies in sub-groups of colorectal cancer.4. Can facilitate regulatory compliance, reduce cost, and accelerate the

development of new therapies

BOTH

Pails

Kras

Braf

PIK3CAPTENAKT

WT/WT

STUDY DESIGN

DNA-based Screening Trial

5-FU/Bev + drug A

5-FU/Bev + drug B

5-FU/Bev + drug C

5-FU/Bev + drug D

5-FU/Bev + drug E

FOLFOX/Bev x 8then

Maintenance

Endpoint PFSN = TBD but likely3000 - 5000

COLON CANCER TASK FORCE OF THENCI GI STEERING COMMITTEE

Progress in Colorectal Cancer

• Subsets of colorectal cancer reliably identifiable with existing technology

• Large-scale efforts needed to bring fruits of this progress to patients

• Public-private partnerships seem most likely to make efficient progress while minimizing cost

• Studies in development in EU, US• Will need support and advocacy by practicing

oncologists and patient groups alike