5 Background: Trastuzumab (H) + capecitabine/fluorouracil + cisplatin is standard of care for 1st-line HER2-positive MGC/GEJC but there is no established HER2-targeted 2nd-line regimen. T-DM1 (H linked to DM1) is approved for HER2-positive metastatic breast cancer previously treated with H + TAX (separately or in combination). This and preclinical HER2-positive GC models provided the rationale for GATSBY (NCT01641939). Methods: GATSBY is a 3-arm randomized, adaptive, seamless phase 2/3 global study of T-DM1 vs TAX in pts with HER2-positive (immunohistochemistry [IHC] 3+ or IHC 2+/in situ hybridization-positive), unresectable LA/MGC/GEJC who progressed during or after 1st-line fluoropyrimidine + platinum ± HER2-targeted therapy. Pts were initially randomized 2:2:1 to T-DM1 3.6 mg/kg every 3 weeks (q3w), T-DM1 2.4 mg/kg weekly (qw), or physician’s choice of paclitaxel 80 mg/m2 qw or docetaxel 75 mg/m2q3w. An independent data monitoring committee selected T-DM1 qw for further study and subsequent patients were randomized 2:1 to this or TAX. The primary endpoint is overall survival (OS); all T-DM1 qw data are analyzed, including dose selection. Results: At clinical cutoff, 06/30/15, 415 pts had been randomized overall: 228 to T-DM1 qw, 117 to TAX (70 T-DM1 q3w pts reported separately); 77.4% had prior HER2-targeted therapy; 29.9% prior gastrectomy; 46.1% were Asian. Efficacy/safety are tabulated. Conclusions: T-DM1 2.4 mg/kg qw did not show an efficacy benefit over TAX. Grade ≥3 AE rates were numerically lower with T-DM1 vs TAX and rates of SAEs, fatal AEs, and treatment discontinuations due to AEs were comparable between arms. Clinical trial information: NCT01641939. [Table: see text]