A Ph1b Study of Demcizumab (DEM, anti-DLL4) with (GEM) in Patients with 1st Line Locally Advanced or Metastatic

1Cubillo A, 2Jameson M, 3Grande E, 4Parnis F, 5Grimison P, 6Cooray P, 7Jeffery M, 8Stagg R, 8Dupont J, 9Tebbutt N 1START, Madrid, Spain; 2Waikato Hospital, Hamilton, New Zealand; 3Ramon y Cajal Hospital, Madrid; Spain; 4Adelaide Cancer Centre, Adelaide, Australia; 5Sydney Cancer Centre, Sydney, Australia; 6Box Hill Hospital, Box Hill, Australia; 7Christchurch Hospital, Christchurch, New Zealand; 8OncoMed, Redwood City, CA; and 9Austin Hospital, Heidelberg, Australia.

Background Patient Demographics (n=24) Pharmacokinetics and Immunogenicity % Change in RECIST Target Lesion Size RECIST Best Overall Response There is accumulating evidence that the cell types within tumors are heterogeneous and Dose Level – mg/kg 2.5* 2.5** 5** Total •To assess the impactof GEM on demcizumabPK, plasma samples were collected pre-and (n=24) that a subset of the cells retain the property to self-renew and give rise to more 40 N88824post-DEM infusion on Study Days 28 and 56 and analyzed for DEM concentration. differentiated progeny. These cells, called Cancer Stem Cells (CSCs) or tumor initiating Dose Level - mg/kg 2.5 2.5 5 Total cells drive tumor growth and metastasis and are more resistant to and Median age (years) 63.5 63.5 69.0 65.5 (16 Evaluable) •Demcizumabclears slowly in patients with a population mean half-life of 16 days. 30 radiotherapy than the remaining tumor cells. The ability to characterize the CSCs through Male/Female 4/4 6/2 3/5 13/11 Partial Response 1 1 2 4 (25%) surface markers and functional limiting tumor dilution assays, using minimally passaged •Gemcitabinedid not appear to significantly alter the PK of demcizumab. human tumors, has enabled the identification of novel agents that specifically target the Prior Surgery 1 2 3 6 20 CSC population. One pathway which appears critical for the CSCs is the Notch pathway. Prior Neoadjuvant/ 10 1 2 StableDisease 4 2 2 7 (44%) The pathway is comprised of 4 Notch receptors (1-4) and 5 ligands, Jagged (1-2) and Adjuvant Therapy Representative Pharmacokinetic Data 10 delta-like ligand (DLL1, 3 and 4). The DLL4 ligand contributes to CSC self-renewal and vascular development. Demcizumab (DEM) is a humanized IgG2 antibody that blocks Prior Radiotherapy 0 0 0 0

or or Size 0 25Q4W2.5Q4W 5Q4W 25Q4W2.5Q4W 25Q4W2.5Q4W 25Q2W2.5Q2W 25Q2W2.5Q2W 25Q2W2.5Q2W 5Q4W 25Q2W2.5Q2W 25Q4W2.5Q4W 5Q4W 25Q4W2.5Q4W 25Q2W2.5Q2W 5Q4W 5Q4W 25Q4W2.5Q4W Clinical Benefit Rate 5 3 4 11 (69%)

DLL4. In minimally passaged human tumor xenografts, DEM was observed to have * Every 2 weeks ** Every 4 weeks m (PR + SD) activity against a variety of tumors including colorectal cancer, breast cancer, lung cancer, pancreatic cancer, and ovarian cancer. The impact of treatment on the -10 frequency of tumorigenicity was assessed using a limiting dilution assay. In several AEs Occurring in >20% of Pts (n=24) Progressive Disease 0 3 2 5 (31%) models, using different chemotherapeutic agents, while the chemotherapy alone -20 decreased tumor volume, the frequency of tumor initiating cells was increased in the All Grades by Dose Level (mg/kg) residual tumor. In contrast, DEM alone decreased the frequency of CSCs and the greatest Not Evaluable 3 2 2 8 Dose Level – mg/kg 2.5 2.5 5 Total (%) % Change in Tu -30 reduction was observed when DEM was combined with chemotherapy. N88824 Nonclinical Xenograft Data -40 Fatigue 5 5 6 16 (67%) Summary Activity of Anti-DLL4 in Combination with Gemcitabine (GEM) in Nausea 6 4 5 15 (63%) -50 Patient Derived Pancreatic Xenograft (OMP-PN4) •Thisis an ongoing Phase 1b dose escalation study of demcizumab,a Vomiting 7 2 6 15 (63%) -60 targeting (targeting the DLL4 GGeyrey lines esepecteda: expected range ge(5 (5th,,50 50th,,95 95th per cen til es) o f co nce nt rat io n-ttepoesbasedoime profiles based on Ph aseapopuatoase 1a population PK aayssanalysis. Reduction in Tumor Volume CSC Frequency Anemia 4 4 6 14 (58%) Symbols: observed individual data in this study by nominal time. st 0.025 ligand in the Notch pathway) plus gemcitabine in 1 line pancreatic 1200 Control Ab

3 Control Ab Thrombocytopenia 2 5 4 11 (46%) Anti-DLL4 0.020 Ant i- DLL4 cancer patients. 1000 • Two of 24 patients developed anti-drug antibodies. Gemcitabine Gemcitabine Decreased appetite 3 3 5 11 (46%) 0.015 800 Gem+Anti-DLL4 Gem+Anti-DLL4 Duration on Study and • Neutralizing antibody analysis is planned at the end of the study. •Demcizumaband gemcitabinewere generally well tolerated with 600 0.010 Peripheral Edema 3 6 2 11 (46%) 400 fatigue and hypertension being the most common drug related CSC Frequency CSC 0.005 Diarrhea 1 5 3 9 (38%) Median Progression Free Survival 200 Tumor Volume, mm Volume, Tumor toxicities. The hypertension was managed with anti-hypertensives. 0.000 0 Abdominal Pain 1 2 5 8 (33%) 30 40 50 60 70 1/137 1/278 1/63 1/833 Reversible Cardiopulmonary Toxicity* Reversible grade 2-3 pulmonary hypertension occurred in 2 patients Constipation 0 3 5 8 (33%) Activity of Anti-DLL4 in Combination with GEM Plus Nab-Paclitaxel in and Grade 2 heart failure occurred in one patient receiving treatment Dyspnea 2 1 4 7 (29%) (Grade 2/3) Patient Derived Pancreatic Xenograft (OMP-PN1) for greater than 100 days. Dose Level- mg/kg 2.5 2.5 5 Total Reduction in Tumor Volume CSC Frequency Hypertension 2 2 3 7 (29%) Che mo+/-mAb (day 0-28) Chemo only •Patients are being followed with cardiac monitoring using B-type 0.07 CSC Frequency Headache 2 2 3 7 (29%) 1200 N

3 0.06 natriuretic pppeptide (()BNP) and echocardiog gpyppraphy. BNP appears to be Control mAb m 1000 Control mAb Neutropenia 0 4 1 5(21%)5 (21%) Gemcitabine 0.05 800 Gem+Abraxane Pulmonary HTN -112 (8%) an early indicator of cardiotoxicity. In addition, a cardioprotective Gem+Abrx 0.04 Anti-DLL4+Gem/Abrx Depression 1 3 1 5 (21%) (Reversible) 600 Anti-DLL4+Gem/Abrx 0.03 medication (i,e, an angiotensin-converting enzyme inhibitor or 400 0.02 carvedilol) was administered to patients with rising BNPs and this CSC Frequency Congestive heart --- - 200 Tumor Volume, m 0.01 failure (Reversible) 0 0.00 strategy appears to prevent cardiotoxicity. 0 20 40 60 Related AEs >15% Pts (n=24) 1/44 Days Post Treatment 1/22 1/554 Right-sided heart --11 (4%) All Grades by Dose Level (mg/kg) failure (Reversible) •Utilizinga truncated treatment approach for demcizumab(i.e. 70days Methods of therapy) appears to prevent the onset of late cardiopulmonary Dose Level – mg/kg 2.5 2.5 5 Total (%) * Occurred following > 100 days of treatment and reversible following the discontinuation of demcizumab and medical management toxicity. This is an open-label Ph 1b dose escalation study of DEM plus GEM in pts with 1st line mPFS pancreatic cancer. The study endpoints included: 1) safety, 2) maximum tolerated dose N88824 Truncated Dosing • 5 mg/kg Q4W: 176 Days •Subsequentcohorts will receive demcizumab in combination with (MTD), 3) immunogenicity, 4) pharmacokinetics (PK), 5) antitumor activity, and 6) • 2.5 mg/kg Q2W: 107 Days gemcitabine and Nab-paclitaxel—this cohort is ongoing. biomarkers of Notch signaling and CSCs. Pts received DEM 2.5mg/kg Q2W or Q4W, or Fatigue 2 3 2 7 (29%) • 2.5 mg/kg Q4W: 50 Days 5mg/kg Q4W with GEM 1000 mg/m2 given 7 of the 1st 8 weeks & then 3 of every 4 weeks • As reversible cardiopulmonary toxicity occurred in 1 patient receiving 5 mg/kg once until disease progression. Dosing of subjects in the1st cohort was paused due to Hypertension 2 2 3 7 (29%) every 4 weeks who was on study for more than 140 days, the subjects in the •Concomitantgemcitabinedid not appear to influence the emergence of reversible cardiotoxicity secondary to DEM in other ongoing studies with subsequent cohorts will receive truncated dosing of DEM (i.e. 70 days of therapy). pharmacokinetics of demcizumab. prolonged treatment of DEM. The protocol was amended to include a cardiac risk Vomiting 3 1 3 7 (29%) mitigation plan including cardiac monitoring with: 1) B-type natriuretic peptide (BNP) and 2) •Three patients in the 5 mg/kg Q4Wcohort had their DEM dosing truncated after this cardiopulmonary event occurred; i.e., their last dose was administered on Day 28, 84 •Four of the 16 (25%) evaluable patients had a RECIST partial echocardiography. Also, 3) cardioprotective medication (i.e, an angiotensin-converting Nausea 3 0 3 6 (25%) enzyme inhibitor or carvedilol) for rising BNP. As reversible cardiopulmonary toxicity and 84, respectively. response and 7 had stable disease. The Kaplan-Meier estimated occurred in 1 patient in the 3rd cohort receiving 5 mg/kg who was dosed for more than 126 Thrombocytopenia 0 3 2 5 (21%) median progression free survival for the 5 mg/kg cohorts was 176 days, the subjects in the subsequent cohorts will receive truncated dosing of DEM (i.e. 70 •They remained on therapy for 204+, 210, and 211 days and none of them developed days. any clinically evident cardiopulmonary toxicity. days of therapy). Subsequent cohorts will also include the administration of Nab-paclitaxel. Peripheral Edema 1 3 1 5 (21%) ADSMBreviewedthedatafromeachdosecohortafterthelastsubjectinthatcohorthad •Two of these patients had a PR and 1 had SD. •A randomized Phase 2 trial in 1st line pancreatic cancer is planned for been on study for 56 days to decide whether it was safe to proceed to the next dose Decreased appetite 2 1 2 5 (21%) The blue bars represent the time on study. T he green bars represent patients who discontinued the study without progression andhave cohort. Data through June 20, 2013 are presented. remained progression-free off study. The patient with the green bar received demcizumab and gemcitabine through day on study 21 and 2014. then received additional gemcitabine off-study 2013 AACR-NCI-EORTC Molecular Targets and Cancer Therapeutics Conference, Boston, 21 October 2013