This phase II trial is studying how well giving irinotecan and cisplatin together with bevacizumab works in treating patients with unresectable or metastatic gastric (stomach) or gastroesophageal junction adenocarcinoma (cancer). Drugs used in chemotherapy, such as irinotecan and cisplatin, work in different ways to stop tumor cells from dividing so they stop growing or die. Monoclonal antibodies such as bevacizumab can locate tumor cells and either kill them or deliver tumor-killing substances to them without harming normal cells. Giving chemotherapy together with a monoclonal antibody may kill more tumor cells.

PRIMARY OBJECTIVES: I. Determine the efficacy of irinotecan, cisplatin, and bevacizumab, in terms of time to progression, in patients with unresectable or metastatic gastric or gastroesophageal junction adenocarcinoma. SECONDARY OBJECTIVES: I. Determine other measures of efficacy, including response rate and median and 1-year survival, in patients treated with this regimen. II. Determine the toxicity of this regimen in these patients. III. Correlate CT perfusion imaging results with the efficacy of this regimen, in terms of time to progression, objective response, and survival, in these patients. IV. Determine the feasibility of serial serum proteomic assays in predicting response to therapy, in terms of time to progression, objective response, and survival, in patients treated with this regimen. V. To bank paraffin stored tumor biopsy material for future planned immunohistochemistry studies to correlate with sensitivity to bevacizumab based combination chemotherapy. OUTLINE: This is an open-label, non-randomized, multicenter study. Patients receive bevacizumab IV over 30-90 minutes on day 1. Patients also receive cisplatin IV over 30 minutes followed by irinotecan IV over 30 minutes on days 1 and 8. Courses repeat every 21 days in the absence of disease progression or unacceptable toxicity. Patients are followed every 3 months for 1 year.