1. A Randomized, Phase 1/2 Trial of AMG 102 or AMG 479 in Combination With Panitumumab vs Panitumumab Alone in Patients With Wild ‑ Type KRAS Metastatic Colorectal Cancer (mCRC): Safety and Efficacy Results Eric Van Cutsem, 1 Cathy Eng, 2 Josep Tabernero, 3 Elzbieta Nowara, 4 Anna Świeboda-Sadlej, 5 Niall C. Tebbutt, 6 Edith P. Mitchell, 7 Irina Davidenko, 8 Lisa Chen, 9 Dominic Smethurst 10 1 University Hospital Gasthuisberg, Leuven, Belgium; 2 The University of Texas M.D. Anderson Cancer Center, Houston, TX; 3 Vall d'Hebron University Hospital, Barcelona, Spain; 4 Instytut im. M. Sklodowskiej-Curie, Gliwice, Poland; 5 Warszawski Uniwersytet Medyczny, Warszawa, Poland; 6 Austin Health, Heidelberg, VIC, Australia; 7 Thomas Jefferson University, Philadelphia, PA; 8 Krasnodar City Oncology Center, Krasnodar, Russia; 9 Amgen Inc., Thousand Oaks, CA; 10 Amgen Ltd., Uxbridge, UK

2. Disclosures Eric Van Cutsem: research funding from Amgen, Merck- Serono, Novartis, Pfizer, Roche, Sanofi-Aventis

3. Introduction Panitumumab, a fully human monoclonal antibody against epidermal growth factor receptor (EGFR), has demonstrated efficacy in patients with wild-type KRAS mCRC in clinical trials 1-4 Rilotumumab (AMG 102) and ganitumab (AMG 479) are investigational, fully human monoclonal antibodies against hepatocyte growth factor (HGF; ligand for c-Met receptor) and insulin ‑ like growth factor 1 receptor (IGF-1R), respectively Preclinical studies indicate that there is complex interdependence between the HGF/c-Met and IGF-1R and EGFR pathways 5-10 Combinations of agents that block these receptors are being investigated for their potential to generate additive/synergistic anticancer effects 1. Van Cutsem E, et al. J Clin Oncol. 2007;25:1658-1664. 2. Amado RG, et al. J Clin Oncol. 2008;26:1626-1634. 3. Peeters M, et al. J Clin Oncol. 2010;28:4706-4713. 4. Douillard JY, et al. J Clin Oncol. 2010;28:4697-4705. 5. Lesko E, et al. Front Biosci. 2008;13:1271-1280. 6. Hynes NE, et al. Nat Rev Cancer. 2005;5:341-354. 7. Jo M, et al. J Biol Chem. 2000;275:8806-8811. 8. Ahmad T, et al. J Biol Chem. 2004;279:1713-1719. 9. Roudabush FL, et al. J Biol Chem. 2000;275:22583-22589. 10. Swantek JL, et al. Endocrinology. 1999;140:3163-3169.

4. Rilotumumab and Ganitumab Mechanisms of Action Ganitumab (AMG 479) targets IGF-1R, inhibiting downstream signaling through PI3K/AKT and MAPK pathways Rilotumumab (AMG 102) targets HGF, inhibiting downstream c-Met signaling Rilotumumab ()

5. Study Schema a Panitumumab 6 mg/kg Q2W; rilotumumab (AMG 102) 10 mg/kg Q2W with dose de-escalation to 5 mg/kg as necessary; primary endpoint was incidence of dose-limiting toxicities b Panitumumab 6 mg/kg Q2W; rilotumumab (AMG 102) dose based on phase 1b; ganitumab (AMG 479) 12 mg/kg Q2W; primary endpoint was ORR c Rilotumumab 10 mg/kg Q2W; ganitumab (AMG 479) 12 mg/kg Q2W; primary endpoint was ORR d Patients in the placebo arm of Part 2 with progressive disease or intolerance to treatment were eligible to participate in Part 3 DLT, dose-limiting toxicity; ORR, objective response rate; Q2W, every 2 weeks Amgen Trial 20060447; ClinicalTrials.gov identifier NCT00788957 Tumor assessments were performed by the investigator using Response Evaluation Criteria in Solid Tumors (RECIST) v1.0

6. Study Objectives Primary Objectives (Part 1 and Part 2) Part 1: To identify a tolerable dose of rilotumumab (AMG 102) in combination with panitumumab based on the incidence and nature of dose-limiting toxicities (DLTs) Part 2: To evaluate the efficacy as measured by the objective response rate (ORR) of rilotumumab (AMG 102) + panitumumab and ganitumab (AMG 479) + panitumumab vs panitumumab + placebo Other Key Objectives (Part 2) Efficacy including progression-free survival (PFS) and overall survival (OS) Safety Pharmacokinetic analysis Biomarker analysis

7. Key Eligibility Criteria Eastern Cooperative Oncology Group (ECOG) performance status score of 0 or 1 Histologically or cytologically confirmed wild-type KRAS mCRC by local or central testing Progression during or following prior treatment with irinotecan- and/or oxaliplatin-based chemotherapy for mCRC Glycosylated hemoglobin ≤ 8% No prior treatment with EGFR, c-Met, or IGF-1R inhibitors

8. Statistical Considerations for Part 2 Randomization was stratified by prior therapy (oxaliplatin or irinotecan vs both) Bayesian analysis of response –This method compares the posterior distribution of the ORR for the experimental arms to that of the control arm to determine an Odds Ratio –An ORR prior distribution for panitumumab monotherapy was derived from 4 previous trials (patients had received both prior oxaliplatin and irinotecan) –The ORR prior distributions for the combination arms were assumed to have the same mean as the panitumumab alone arm –The ORR posterior distribution for each arm combines the prior distributions with observed ORRs from the study It was prespecified that if there was ≥ 90% probability that combination therapy was better than panitumumab alone as evaluated by objective tumor response, the combination was considered promising –If there was between 50% and 90% probability, the combination was considered indeterminate –If there was < 50% probability, the combination was considered not promising

9. Results Part 1 Results (Phase 1b) In Part 1, no DLTs were reported for the first 6 DLT- evaluable patients receiving panitumumab in combination with rilotumumab (AMG 102) 10 mg/kg Q2W –The 10 mg/kg Q2W dose of rilotumumab (AMG 102) was used in Part 2 Part 2 Results (Phase 2) 142 patients enrolled from 37 sites in 11 countries The enrollment period was June 9, 2009 through February 5, 2010 The date for data cut-off for this analysis was July 23, 2010 Median follow-up is 6.9 months; follow-up is ongoing

10. Part 2: Patient Demographics and Disease Characteristics at Baseline Panitumumab + Placebo (n = 48) Panitumumab + Rilotumumab (AMG 102) (n = 48) Panitumumab + Ganitumab (AMG 479) (n = 46) Men - n (%)28 (58)29 (60)25 (54) Age - mean years (range)55.0 (19-75)62.1 (45-78)62.0 (33-81) ECOG status - n (%) 0 1 15 (31) 33 (69) 24 (50) 23 (48) a 18 (39) 28 (61) Metastatic sites - n (%) Liver only Liver + other sites 5 (10) 27 (56) 5 (10) 32 (67) 4 (9) 29 (63) Prior therapies for mCRC - n (%) First-line therapy Second-line therapy Third-line therapy and later 46 (96) b 31 (65) 14 (29) 48 (100) 33 (69) 16 (33) 46 (100) 26 (57) 12 (26) Prior chemotherapies for mCRC - n (%) Oxaliplatin Irinotecan Oxaliplatin and irinotecan 39 (81) 30 (63) 23 (48) 42 (88) 32 (67) 26 (54) 40 (87) 26 (57) 20 (44) a One patient with ECOG performance score of 2 was enrolled in error; data from this patient were included in all efficacy and safety analyses b Two patients had not received first-line therapy for mCRC; both patients had received oxaliplatin-based chemotherapy for non-metastatic CRC in the adjuvant setting and progressed on therapy before entering the study

11. Part 2: Primary Endpoint Overall Response Rate Panitumumab + Placebo (n = 48) Panitumumab + Rilotumumab (AMG 102) (n = 48) Panitumumab + Ganitumab (AMG 479) (n = 46) Objective Response - n (%) Complete Response (CR) Partial Response (PR) Stable Disease (SD) Progressive Disease (PD) Unevaluable/Not done 10 (21) 0 (0) 10 (21) 17 (35) 16 (33) 5 (10) 15 (31) 0 (0) 15 (31) 19 (40) 11 (23) 3 (6) 10 (22) 0 (0) 10 (22) 18 (39) 15 (33) 3 (6) Disease control rate a - % (95% CI)56 (41-71)71 (56-83)61 (45-75) Duration of response - median months (95% CI) 3.7 (3.6-NE)5.1 (3.7-5.6)3.7 (3.6-5.8) Posterior probability of Odds Ratio > 1 b 0.930.63 a Disease control rate = CR + PR + SD b OR is calculated based on ORR; an OR > 1 favors the combination arm over panitumumab alone NE, not estimable Responses were required to be confirmed at least 4 weeks after response criteria were first met

12. Part 2: Progression-Free Survival Panitumumab ± Rilotumumab (AMG 102) Panitumumab ± Ganitumab (AMG 479) (AMG 102) (AMG 479) (AMG 102) (AMG 479)

13. Adverse Events in Part 2 (Any Grade in ≥ 20% or Grade 3/4 in ≥ 2 Patients) Panitumumab + Placebo (n = 48) Panitumumab + Rilotumumab (AMG 102) (n = 48) Panitumumab + Ganitumab (AMG 479) (n = 46) AE (Preferred term) - %Any GradeGrade 3/4Any GradeGrade 3/4Any GradeGrade 3/4 Any AE9452987110063 Rash52858294813 Acneiform dermatitis331035152611 Pruritus250210282 Skin fissures170152260 Paronychia152314202 Dry skin150232220 Acne0084110 Skin toxicity002244 Constipation256100130 Decreased appetite172212202 Abdominal pain15610497 Diarrhea100154262 Hypomagnesemia2122944115 Fatigue212104172 Anemia1784020 Asthenia15080134 There were 9 grade 5 AEs; 1 occurred in the panitumumab alone arm and 4 occurred each in the combination arms ― All except 1 were due to disease progression; 1 fatal AE was due to staphylococcal sepsis (panitumumab + ganitumab [AMG 479] arm) ― None were reported to be related to investigational product AE, adverse event

14. Conclusions This is the first study to show promising evidence of efficacy by an HGF (c-Met pathway) inhibitor (rilotumumab [AMG 102]) when combined with panitumumab in patients with mCRC The activity as assessed by ORR for patients receiving rilotumumab (AMG 102) plus panitumumab is promising (per prospectively specified Bayesian criterion) Efficacy of ganitumab (AMG 479) plus panitumumab combination therapy as determined by ORR was indeterminate The safety profiles of the drug combinations were generally similar to that of panitumumab alone with some exceptions, including a higher rate of grade 3/4 rash with rilotumumab and of hypomagnesemia with ganitumab Analyses of biomarkers of response in serum and tissue samples are underway