Background
Whether neoadjuvant DOS (docetaxel/oxaliplatin/S-1) improves long-term outcomes compared with SOX (oxaliplatin/S-1) in HER2-negative LAGC remains uncertain, particularly in routine practice.
Methods
This multicenter retrospective cohort included consecutive HER2-negative cT2N⁺ or cT3–4NanyM0 gastric adenocarcinoma patients receiving ≥2 cycles of neoadjuvant DOS or SOX (four Chinese tertiary centers, 2010–2024). Disease-free survival (DFS; surgery to relapse/death) and overall survival (OS; surgery to death) were secondary endpoints. Survival analyses used Kaplan–Meier and Cox proportional hazards models after PSM (1:1) and IPTW. A perioperative-consistency sensitivity cohort (same regimen pre- and post-operatively) was included and MPR–survival association analyses were performed. Prespecified subgroup analyses assessed effect heterogeneity.
Results
A total of 1,283 patients were analyzed (DOS 461; SOX 822). DFS analyses were performed in R0-resected radical surgery patients (median follow-up 35.1 months; 430 DFS events), and OS analyses in the entire cohort (median follow-up 34.9 months; 519 deaths). In unadjusted DFS comparisons, DOS did not significantly outperform SOX; however, adjusted analyses showed favorable trends (PSM HR 0.82, 95% CI 0.66–1.02; IPTW HR 0.82, 95% CI 0.66–1.02). In multivariable Cox models, DOS significantly reduced progression risk in IPTW (HR 0.74, 95% CI 0.59–0.93; P=0.010) and was borderline in PSM (HR 0.82, 95% CI 0.65–1.00; P=0.052). OS was similar across regimens in all cohorts. Findings were consistent in the perioperative-consistency sensitivity cohort. Achieving MPR was associated with longer DFS (HR 0.64, 95% CI 0.50–0.82) and OS (HR 0.72, 95% CI 0.56–0.92). DFS benefit of DOS was most evident in patients aged ≥65 years, those with lower tumor burden (cT2–3 or cN0–1), and proximal tumors.
Conclusions
Neoadjuvant DOS improved progression control versus SOX after rigorous adjustment without an OS difference at current follow-up; MPR strongly correlated with survival, and selected subgroups may derive greater DFS benefit.
