Background
Phase II evidence suggests that adding docetaxel to S-1/oxaliplatin may deepen pathological response in HER2-negative locally advanced gastric cancer (LAGC), but large head-to-head real-world evidence versus the standard SOX regimen is limited.

Methods
We retrospectively reviewed consecutive HER2-negative gastric adenocarcinoma patients (cT2N⁺ or cT3–4NanyM0) treated with ≥2 cycles of neoadjuvant DOS (docetaxel/oxaliplatin/S-1) or SOX (oxaliplatin/S-1) at four high-volume Chinese centers (2010–2024). Major pathological response (MPR; Mandard TRG 1–2) was the primary endpoint; perioperative outcomes included R0 resection and severe postoperative complications (Clavien–Dindo ≥III). Confounding was addressed using propensity-score matching (PSM, 1:1) and inverse probability-of-treatment weighting (IPTW). Robustness was assessed by effect-size metrics (OR/RR) and E-values.

Results
Among 1,283 eligible patients (DOS 461; SOX 822), PSM yielded 446 well-balanced pairs (all SMD<0.10), and IPTW achieved good balance with stabilized/truncated weights. DOS consistently improved MPR versus SOX across analytic cohorts: unadjusted, 25.7% vs 17.8% (P=0.002); PSM, 25.7% vs 17.8% (P=0.009); and IPTW, 24.8% vs 17.4% (P=0.005). In the matched cohort, DOS increased MPR (OR 1.59, 95% CI 1.14–2.23; RR 1.44, 95% CI 1.10–1.88), with E-values >1.8 across cohorts. Surgical quality and safety were comparable: R0 rates >95% among radical gastrectomy patients, severe complications were rare and identical in the PSM cohort (2.2% vs 2.2%), and perioperative mortality was <1% in both groups.

Conclusions
In a large multicenter real-world cohort, neoadjuvant DOS achieved higher MPR than SOX without compromising perioperative safety, supporting taxane-based triplets as an effective perioperative option for fit HER2-negative LAGC patients.