Background: To explore proper agents for AGC patients as 2nd-line treatment based on optimal biomarker, we conducted K-Umbrella GC study with standard of care (SOC) controlled umbrella trial design. Methods: HER2-negative AGC patients from 6 Korean cancer centers were centrally screened for druggable targets by IHC and in situ hybridization. Patients were randomized to the biomarker vs. control group with SOC as 4:1 ratio. In the biomarker group, patients were treated with specific targeted agents in combination with weekly paclitaxel; 1) EGFR 2+/3+ patients for pan-ERBB inhibitor (afatinib; EGFR cohort), 2) PTEN loss/null (H-score <100) patients for PIK3Cβ inhibitor (GSK2636771; PTEN cohort), and 3) PD-L1+, dMMR/MSI-high, or EBV-related cases for anti-PD-1 inhibitor (nivolumab; NIVO cohort). Control group and NONE cohort in biomarker group without predefined biomarkers were treated with SOC (weekly Paclitaxel±Ramucirumab). Primary endpoint was PFS between control and biomarker groups. Secondary endpoints included efficacy and safety of each cohort. Results: Between Feb 2016 and Feb 2021, total 722 patients were centrally screened. 329 patients were enrolled and randomized to control group (n=63) or biomarker group (n=266; EGFR cohort n=67; PTEN cohort n=42; NIVO cohort n=54; NONE cohort n=103). With a median follow-up of 35 months (95%CI 26.1-55.3), median PFS and OS were 3.8 (95%CI 3.2-4.3) and 8.9 (95%CI 7.8-10.1) months for biomarker group and 4.1 (95%CI 3.0-4.8) and 8.7 months (95%CI 7.2-10.2) for control group. In control group, PTEN loss/null was poor prognostic marker; patients with PTEN loss/null (n=12) showed worse survival compared to PTEN intact patients (n=51) (mPFS, 2.8 vs 4.3 months, P=0.03; mOS, 8.7 vs 9.1 months), where other biomarkers showed similar prognosis. Afatinib for EGFR cohort or GSK2636771 for PTEN cohort did not show significant survival benefit compared to control group (Table). Among patients with immune-related biomarkers, addition of nivolumab showed durable survival benefit (mOS 12.0 vs 7.6 months, P=0.08) compared to SOC. Conclusions: Considering the characteristics of umbrella design with multiple biomarkers having different biological roles, biomarker group did not show the improved survival over control arm with these 3 drugs. For optimal biomarker-driven targeted agent applications, NGS-based biomarker driven K-Umbrella GC-2 study is ongoing. Clinical trial information: NCT02951091.