MAIC of Fruquintinib Plus Paclitaxel Versus Ramucirumab Plus Paclitaxel in Advanced G/GEJ Adenocarcinoma

This retrospective MAIC analysis employs individual patient data (IPD) from the FRUTIGA trial (fruquintinib arm) and published aggregate data (AgD) from RAINBOW-Asia (ramucirumab arm), with placebo as the common anchor. Pseudo individual participant data (Pseudo-IPD) for the RAINBOW-Asia trial were reconstructed from published Kaplan-Meier curves using the Guyot algorithm (2012).

•Weighting Methodology: Seven prognostic factors balanced: age <65, male sex, ECOG 0, GEJ primary, peritoneal metastases, metastatic sites, prior doublet chemotherapy Optimization via BFGS algorithm (convergence tolerance 1e-6) Effective sample size (ESS) retention: > 50%

•Statistical Analysis: Primary: Adjusted PFS hazard ratio (HR) using Bucher method with 95% bootstrap CI (100 iterations) Secondary: Weighted Cox models for OS; logistic regression for ORR/DCR Sensitivity: Simulated Treatment Comparison (STC) and covariate threshold analyses

•Sensitivity Analyses: RMST analyses were conducted as supportive evidence alongside primary Cox models for time-to-event endpoints violating proportional hazards assumptions.

Restricted mean survival time (RMST) Simulated Treatment Comparison (STC) Bootstrap confidence intervals (100 iterations)

Inclusion criteria

• Histologically confirmed gastric/GEJ adenocarcinoma
• Advanced or metastatic disease
• ECOG 0-1
• Received either fruquintinib + paclitaxel or reference regimen
• Available baseline characteristics for matching variables

Exclusion criteria

• Missing key outcome data
• Incomplete baseline characteristics for >2 matching variables
• Prior fruquintinib exposure (control arm only)