221P - Evaluation of MET staining in gastric/gastroesophageal junction (G/GEJ) tumor samples as a biomarker for rilotumumab (R) benefit

Date 28 September 2014
Event ESMO 2014
Session Poster Display session
Topics Oesophageal Cancer
Gastric Cancer
Translational Research
Presenter Michael Hale
Citation Annals of Oncology (2014) 25 (suppl_4): iv58-iv84. 10.1093/annonc/mdu326
Authors M.D. Hale1, K.S. Oliner2, R. Tang1, J.G. Vallone3, I. Klement4, S. Webster5, L. Chen1, E. Loh6, S.D. Patterson2
  • 1Biostatistics, Amgen Inc., 91320 - Thousand Oaks/US
  • 2Medical Sciences - In Vitro Diagnostics, Amgen Inc., Thousand Oaks/US
  • 3Clinical Pathology, University of Southern California Keck School of Medicine, Los Angeles/US
  • 4Pathology, Cottage Hospital, Santa Barbara/US
  • 5Companion Diagnostics R&d, Dako North America, an Agilent Technologies Company, Carpinteria/US
  • 6Oncology, Amgen Inc., South San Francisco/US



R is an investigational, fully human monoclonal antibody against hepatocyte growth factor, the only known ligand for the MET proto-oncogene. In a phase 2 study in G/GEJ cancer, trends towards improved overall survival (OS) and progression-free survival (PFS) were seen with R + epirubicin, cisplatin, and capecitabine (ECX) vs placebo + ECX, with benefit due to the MET-positive patients (pts). We describe a methodology to select a cutoff for defining MET positivity.


Eligible pts had unresectable locally advanced or metastatic G/GEJ adenocarcinoma, ECOG performance status ≤ 1, and no prior systemic therapy for this disease. Pts were randomized 1:1:1 to R 15 mg/kg, R 7.5 mg/kg, or placebo IV day 1 Q3W plus ECX (50 mg/m2 IV day 1, 60 mg/m2 IV day 1, 625 mg/m2 BID orally days 1–21, respectively). Formalin-fixed paraffin-embedded archival tumor tissues were stained with the Dako MET IHC pharmDx™ kit, which uses the MET4 antibody. Membrane, cytoplasmic, and total (cytoplasmic and membrane) tumor cell staining were evaluated separately. Percent staining at different intensities (0, 1 + , 2 + , 3+) and combinations of intensities were scored directly, whereas overall percent positive, H-score, and predominant staining intensity were derived scores. OS and PFS for subgroups defined by potential cutoff values (5–95%) at 5% increments were analyzed by Cox proportional hazards regression, Kaplan-Meier, and the log rank test, yielding hazard ratios (HRs), median OS and PFS, and p values.


121 pts were randomized; 91 had tumor samples evaluable for MET IHC. Pts whose tumors showed ≥ 25% membranous staining were classified as MET-positive. A strong treatment benefit (R vs placebo) was seen in MET-positive pts (n = 58; OS: HR = 0.46, 95% CI, 0.24–0.87; PFS: HR = 0.46, 95% CI, 0.25–0.85); median OS: 10.6 vs 5.7 mo; median PFS: 6.8 vs 4.4 mo. No benefit or detriment was seen with R in MET-negative pts (n = 33; OS: HR = 1.23; 95% CI, 0.56–2.70; PFS: HR = 1.00; 95% CI, 0.46–2.16). Cutoff values 25–50% showed a similar benefit of R in MET-positive pts.


The subgroup defined by a cutoff of 25% MET-positive membranous staining using the Dako MET IHC pharmDx™ kit showed benefit from R + ECX; no negative impact of R was seen in MET-negative pts. This cutoff is being used to select pts in an ongoing phase 3 study of R + ECX in MET-positive G/GEJ cancer (NCT01697072).


M.D. Hale, K.S. Oliner, R. Tang, L. Chen, E. Loh and S.D. Patterson have declared:is an employee of and owns stock in Amgen Inc.; S. Webster: Scott Webster is an employee of Dako North America, an Agilent Technologies Company. All other authors have declared no conflicts of interest.