Phase 2 Study of AMG 386 (20060439) in combination with Cisplatin & Capecitabine in Subjects with Metastatic Gastric, Gastroesophageal Junction, or Distal Esophageal Adenocarcinoma

- Disease Related •Histologically or cytologically confirmed adenocarcinoma of the stomach, gastroesophageal junction or distal esophagus with metastatic disease •Measurable or non-measurable disease per RECIST Guidelines •Prior gastrectomy (total or partial) may be allowed as long as subjects can take oral medications and meet all other inclusion/exclusion criteria. Subjects may not take crushed or dissolved capecitabine via a feeding/gastrostomy tube •Palliative radiotherapy for metastatic esophageal or gastric cancer prior to study entry may be allowed as long as all toxicities from radiotherapy have resolved and the radiotherapy was not to the only site of known metastatic disease - Demographic •18 years of age or older at the time the written informed consent is obtained - General •Able to swallow oral medication •ECOG performance status of 0 or 1 - Laboratory •Adequate organ and hematological function as evidenced by laboratory studies prior to randomization

- Disease Related •Prior chemotherapy for metastatic disease (1st line) •Less than 12 months have elapsed from completion of previous adjuvant or neoadjuvant chemotherapy or chemoradiotherapy •Subjects with persistent gastric outlet obstruction, complete dysphagia or feeding jejunostomy • Radiotherapy ≤ 14 days prior to randomization. Subjects must have recovered from all radiotherapy-related toxicities •Current or prior history of central nervous system metastases •History of arterial or deep venous thromboembolism within 12 months prior to randomization •History of bleeding diathesis or clinically significant bleeding within 6 months prior to randomization •Major surgical procedure within 28 days prior to randomization •Minor surgical procedure, placement of central venous access device or fine needle aspiration within 3 days prior to randomization •Prior malignancy except: malignancy treated with curative intent and without evidence of active disease for ≥ 3 years prior to randomization and felt to be at low risk for recurrence by treating physician, adequately treated non-melanomatous skin cancer or lentigo maligna without evidence of disease, adequately treated cervical carcinoma in situ without evidence of disease, prostatic intraepithelial neoplasia without evidence of prostate cancer •Prior malignancy (other than in situ cervical cancer, or basal cell cancer of the skin) unless treated with curative intent and without evidence of disease for ≥ 3 years prior to randomization •Clinically significant cardiovascular diseases within 12 months prior to randomization •Presence of clinically significant non-healing wound, ulcer (including gastrointestinal) or fracture as judged by the investigator •Ongoing or clinically significant active infection as judged by the investigator •Known hypersensitivity to bacterial proteins, or any of the drugs required in this study •Known peripheral neuropathy ≥Grade 1 •Known dihydropyrimidine dehydrogenase deficiency •Known hypersensitivity to 5-FU/capecitabine •Known positive test for human immunodeficiency virus (HIV), hepatitis C, or hepatitis B surface antigen •Known active or chronic hepatitis - Medications •Currently or previously treated with angiopoietin inhibitors, or inhibitors of TIE-1 or TIE-2 •Treatment with immune modulators such as cyclosporine or tacrolimus within 30 days prior to randomization •Treatment with sorivudine or its chemically related analogues • Anticoagulants (other than aspirin and anti-platelet agents) within 7 days prior to randomization. The concurrent use of low molecular weight heparin or heparanoids or low dose warfarin (i.e, ≤ 1 mg daily) for prophylaxis against thrombosis is acceptable while on study - General •Not yet completed at least 30 days since ending other investigational device/drug trial(s), or subject is receiving other investigational treatments •Pregnant or is breast feeding