Biomarker defined subgroups in gastric cancer, a case study

Nigel Baker

EFSPI, November 30, 2012

Overview

• Background• Gastric cancer• Biomarkers

• Phase 2 trial• Overall• Biomarker subgroup

• Conclusions

• Lessons

2

Gastric cancer

• Second leading cause of cancer-related death

• Poor 5-year survival (~25%) all stages combined

• ~10 month survival for metastatic stage

• Late diagnosis with no standard screening in North America: ~60% regional or metastatic disease at diagnosis

• No single global standard of care chemotherapy: epirubicin, cisplatin, fluoropyrimidine, based regimens most commonly used

3

Stomach - AnatomyStomach - Anatomy

Possible markers to define subgroups

© www.Abcam.com

PIPPIP33PIPPIP22

Gab1Gab1PI3KPI3K

ERKERK

MEKMEKRasRas

Shp2Shp2Gab1Gab1

METMET

ProliferationProliferationMigrationMigrationInvasionInvasion

cdc42cdc42Rac1Rac1

RafRafAKTAKT

SurvivalSurvival

HGF/SFHGF/SF

Rilotumumab (AMG 102)Rilotumumab (AMG 102)

X XX

Possible markers to define subgroups

• Pathway defined potential biomarkers

5

HGF/MET Pathway in Gastric Cancer

• The MET receptor and its ligand, hepatocyte growth factor (HGF), also known as scatter factor (SF), regulate multiple cellular functions, including proliferation, survival, motility, and morphogenesis1

• MET and HGF are important in gastric and esophagogastric junction (G/EGJ) cancer

• MET is expressed in 26% to 74% of gastric cancer (GC) cases, and MET is amplified in 2% to 23% of GC cases1-8

• MET expression is associated with tumor depth of invasion, metastasis, stage, and poor prognosis in patients with GC2,3,6,7,9

• MET expression occurs in 80% to 100% of EGJ cases and is associated with poor outcome10,11

• Biomarkers can potentially be used to identify patients likely to benefit from agents targeting the HGF/MET pathway

• Rilotumumab is an investigational, fully human monoclonal antibody to HGF that blocks the binding of HGF to MET

Key Inclusion Criteria•Pathologically confirmed unresectable locally advanced or metastatic G/EGJ cancer•No prior systemic therapy for locally advanced or metastatic disease

Endpoints•Primary: PFS•Others: OS, ORR, PK, biomarkers

Patient Enrollment/Sites • 121 patients were randomized from 42

sites (from 13 countries)

Data Analyses• Estimation study to evaluate efficacy endpoints

– Each dose of rilotumumab separately– Combined rilotumumab group

• Primary analysis: 79 PFS events, November 30, 2010• Updated analysis:102 PFS events and 74 OS events (~ 60%), April 1, 2011

ARM A (n=40)Rilotumumab (15 mg/kg) + ECX Q3W

ARM B (n=42)Rilotumumab (7.5 mg/kg) + ECX Q3W

ARM C (n=39)Placebo+ ECX Q3W

E: Epirubicin: 50 mg/m2 IV, day 1C: Cisplatin: 60 mg/m2 IV, day 1

X: Capecitabine: 625 mg/m2 BID orally, days 1-21

Stratification factors:ECOG PS 0 vs 1LA vs metastatic

Phase 2 Study of Rilotumumab + Epirubicin, Cisplatin, and Capecitabine (ECX) in G/EGJ cancer

0

20

40

60

80

100

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Pro

gres

sion

-Fre

e Su

rviv

al (%

)

Time (months)

0

20

40

60

80

100

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17Time (months)

Ove

rall

Surv

ival

(%)

Clinical Efficacy in the Intent-to-Treat Population*

*Results based on the updated analysis with data cutoff of April 1, 2011. †Adjusted for baseline randomization stratification variables (ECOG status [0 or 1] and disease extent [locally advanced or metastatic]).Iveson T, et al. European Multidisciplinary Cancer Congress, September 23-27, 2011, Stockholm, Sweden; abstract #6504.

Median Months(80% CI)

HR†

(80% CI) P Value

Rilotumumab + ECX (n = 82) 5.6 (4.9–6.9) 0.64 (0.48–0.85) 0.045

Placebo + ECX (n = 39) 4.2 (3.7–4.6)

Progression-Free Survival

Overall Survival Median Months(80% CI)

HR†

(80% CI) P Value

Rilotumumab + ECX (n = 82) 11.1 (9.5–12.1) 0.73 (0.53–1.01) 0.215

Placebo + ECX (n = 39) 8.9 (5.7–10.6)

Hypothesis testing requirements:Patient stratification biomarkers• Effective drug or evidence of biologic activity

• Well-defined biologic hypothesis• Knowledge of pathway and target marker• Meaningful biomarker variation that is prevalent in indication of interest

(e.g., KRAS in colorectal vs head and neck cancer)

• Appropriate setting to perform robust test• Adequate sample size with high rate of marker ascertainment

• Depends on both biomarker prevalence and level of efficacy• Control arm to dissect prognostic vs. predictive value of marker

• Validated assay that addresses testable hypothesis• KRAS: limited known mutations, DNA assay• PTEN: huge gene, cytoplasmic and nuclear protein

challenges in assay interpretation and standardization

3

Biomarker Enrichment?

• Primary analysis – encouraging

• Strong historical evidence for a predictive and prognostic biomarker

• Pre-specified analysis plan to find biomarker that defines enriched population

• Aims:• Continue Amgen’s focus on personalised medicine• Strengthen chance of surviving portfolio planning• Strengthen chance of Regulatory success

10

Sample Ascertainment

Patients,* n (%)Patients, n 118

MET immunohistochemistryAcceptable tumor sample available, n (%) 90 (76%)

†METHigh, n (%) 38 (42%)†METLow, n (%) 52 (58%)

Evaluable patients in each treatment arm Rilotumumab + ECX 62 (78%)

Placebo + ECX 28 (74%)

*Per protocol analysis set.

†Criteria chosen for clinical trial assay:METHigh:> 50% of tumor cells with cytoplasmic stainingMETLow: ≤ 50% of tumor cells with cytoplasmic staining

Choosing Cut Point - OS by MET Expression With Increasing Threshold of Percent Positive

• Cox PH modeling suggests increasingly

favorable HR up to a 50% cutpoint.

– Above 50%, no apparent gain is observed.

BiomarkerSubgroup

(% cell staining) HR (95% CI) P valueInteraction

P value

FavorsRilotumumab + ECX

FavorsPlacebo + ECX

0.001 0.01 0.1 1.0 10.0

High (> 50) 0.290 (0.111–0.760) 0.012 0.007Low (≤ 50) 1.838 (0.778–4.343) 0.166High (> 40) 0.501 (0.228–1.098) 0.084 0.051Low (≤ 40) 1.699 (0.669–4.312) 0.265High (> 30) 0.710 (0.327–1.541) 0.387 0.332Low (≤ 30) 1.286 (0.498–3.323) 0.604High (> 20) 0.855 (0.414–1.765) 0.672 0.769Low (≤ 20) 1.027 (0.360–2.933) 0.960High (> 10) 0.847 (0.429–1.672) 0.633 0.563Low (≤ 10) 1.469 (0.406–5.313) 0.557

High (> 90) 0.035 (0.002–0.678) 0.027 0.028Low (≤ 90) 1.209 (0.625–2.337) 0.573High (> 80) 0.166 (0.033–0.823) 0.028 0.010Low (≤ 80) 1.473 (0.700–3.102) 0.308High (> 70) 0.292 (0.099–0.858) 0.025 0.012Low (≤ 70) 1.641 (0.743–3.628) 0.221High (> 60) 0.262 (0.095–0.725) 0.010 0.004Low (≤ 60) 1.879 (0.806–4.381) 0.144

0

20

40

60

80

100

Time (Months)0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Ove

rall

Surv

ival

(%)

PFS and OS May be Improved in METHigh

PatientsProgression-Free Survival

Overall Survival

*HR adjusted for baseline disease extent and ECOG PS.

Time (Months)0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

100

80

60

40

20

0

Prog

ress

ion-

Free

Surv

ival

(%) Median Months

(80% CI)HR*

(95% CI) P Value

Rilotumumab + ECX (n = 27) 6.9 (5.1–7.5) 0.51 (0.24–1.10) 0.085

Placebo + ECX (n = 11) 4.6 (3.7–5.2)

Median Months(80% CI)

HR+

(95% CI) P Value

Rilotumumab + ECX (n = 27) 11.1 (9.2–13.3) 0.29 (0.11–0.76) 0.012

Placebo + ECX (n = 11) 5.7 (4.5–10.4)

Distribution of Bootstrap Hazard Ratios

14

0.64

Percentile HR

100 2.43

95 0.64

50 0.26

5 0.07

0 0

Ove

rall

Surv

ival

(%)

0102030405060708090

100

Time (Months)0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

High Levels of Tumor MET May Be Prognostic and Predictive

• Patients with gastric tumors with high MET expression may have a poorer prognosis but may receive more benefit from rilotumumab

Median Months(80% CI)

Logrank P value

Placebo + ECX (METLow, n = 17) NE (8.5–NE)0.023Placebo + ECX (METHigh, n =

11) 5.7 (4.5–10.4)

0.01 0.1 1.0 10.0 100.0

Favors METHigh Favors METLow

Rilotumumab + ECX 0.677 (0.350, 1.310) 0.2470.007

Placebo + ECX 3.218 (1.076, 9.630) 0.037

TreatmentGroup HR (95% CI) P value

InteractionP valueTreatment Group HR (High vs Low)

(95% CI) P-value Interaction p-value

Rilotumumab + ECX 0.68 (0.35, 1.31) 0.2470.023

Placebo + ECX 3.22 (1.08, 9.63) 0.037

Rilotumumab Safety Profile in the METHigh

Patient Subgroup

METHigh METLow Unselected†

Rilotumumab(n = 27)

Placebo(n = 11)

Rilotumumab(n = 35)

Placebo(n = 17)

Rilotumumab(n = 81)

Placebo(n =39)

Any AE, n (%) 27 (100) 11 (100) 34 (97) 17 (100) 80 (99) 39 (100)

Grade ≥ 3 AE, n (%) 24 (89) 9 (82) 31 (89) 13 (76) 70 (86) 29 (74)

Serious AE, n (%) 16 (59) 7 (64) 22 (63) 6 (35) 47 (58) 19 (49)

Fatal AE, n (%) 0 (0) 2 (18) 8 (23) 2 (12) 9 (11) 6 (15)

†Safety analysis set.

Conclusions

• High tumor MET expression may predict clinical benefit for the addition of rilotumumab to ECX in patients with advanced G/EGJ cancer

• The outcome by MET expression in the placebo arm suggests that high tumor levels of MET is a marker of poor prognosis that may be reversed by rilotumumab

• Rilotumumab has a manageable safety profile among patients with high tumor MET expression

• Based on these results, a phase 3 trial is planned to confirm the efficacy of rilotumumab added to ECX in patients with MET-positive G/EGJ tumors as detected by a MET in vitro diagnostic assay

Lessons

• May increase probability of success• Internal decision making• Predicted Regulatory success

• May make Regulatory buy-in to a Phase 3 design easier• Exclude patients who are less likely to benefit

• Needs careful assessment• Biological justification• Pre-specified analysis plan

18

Acknowledgements

• Kelly Oliner, Amgen for her ASCO 2012 Slides

• Elwyn Loh, Rui (Sammi)Tang,Joe Jiang and Chrissie Fletcher for helpful review comments

19

References

• Slide 3:Jemal et al, CA Cancer J Clin 2011;61: 60-90Jemal et al, CA Cancer J Clin 2010;60;277-300Kang et al, Curr Treat Options Oncol 2011;12:96-106

• Slide 6:

• Slide 9:2008 ODAC briefing book

Altar CA, Amakye D, Bounos D, et al. A prototypical process for creating evidentiary standards for biomarkers and diagnostics.

Clin Pharmacol Ther. 2008;83:368-71.

Hayes DF, Bast RC, Desch CE, et al. Tumor marker utility grading system: a framework to evaluate clinical utility of tumor markers.

J Natl Cancer Inst. 1996;88:1456- 66.

Woolsey R. Drug-diagnostic co-development: a new paradigm. 2008; http://www.cpath.org/Portals/0/PersonMedv2.pdf

Food and Drug Administration. Drug-diagnostic co-development concept paper. 2005; http://www.fda.gov/Cder/genomics/pharmacoconceptfn.pdf.

20

Nakajima M, et al. Cancer. 1999;85:1894-1902.Taniguchi T, et al. Br J Cancer. 1997;75:673-677.Wu C, et al. Oncol Rep. 1998;5:817-822.Graziano F, et al. J Clin Oncol. 2011;29:4789-4795.Tuynaman JB, et al. Br J Cancer. 2008;6:1102-1108.

Birchmeier C, et al. Nat Rev Mol Cell Biol. 2003;4:915-925.Janjigian YY, et al. Cancer Epidemiol Biomarkers Prev. 011;20:1021-1027.Lennerz JK, et al. J Clin Oncol. 2011;29:1-8.Drebber UTA, et al. Oncol Rep. 2008;19:1477-1483.Amemiya H, et al. Oncol. 2002;63:286-296.Kubicka S, et al. Dig Dis Sci. 2002;47:114-121.