Prediction of perioperative FLOT outcomes in Resectable Gastric and GEJ Adenocarcinoma with pathologic and molecular features: A Retrospective Real-World Cohort Study.

BACKGROUND: Perioperative FLOT is the reference regimen for resectable gastric and gastroesophageal junction (GEJ) adenocarcinoma, yet predictors of response and survival remain incompletely defined. We evaluated pathological response, molecular associations, and survival in a real-world cohort from a high-volume Italian institution, with the aim of benchmarking contemporary outcomes and generating hypotheses for future risk stratification.

METHODS: Consecutive patients with stage IB-IVa resectable GC/GEJ treated with neoadjuvant FLOT (2017-2025) at a tertiary center were retrospectively analyzed. Primary endpoints were Mandard tumor regression grade (TRG) and overall survival (OS) from diagnosis. Secondary endpoints included event-free survival (EFS), disease-free survival (DFS), R0 rate, safety, and associations between clinicopathologic/molecular features (HER2, PD-L1 CPS, MMR) and TRG. Multivariable logistic and Cox models were applied.

RESULTS: Among 159 patients, 95% proceeded to resection; 11.9% progressed during neoadjuvant therapy. Pathological complete response was 6.9%, major response (TRG1-2) 17.0%, and R0 rate 92% in non-metastatic resections. Severe (≥G3) neoadjuvant toxicity occurred in 15%. Two-year OS and EFS from diagnosis were 78.2% and 58%, respectively; two-year DFS after R0 resection was 64.5%. Signet-ring cells and clinical nodal positivity independently predicted poorer TRG. PD-L1 CPS>1 was associated with inferior tumor regression in exploratory analysis. Baseline cT, cN, and cM stage predicted EFS, whereas post-therapy stage ypTNM and receipt of adjuvant therapy were associated with DFS.

CONCLUSIONS: In real-world practice, perioperative FLOT shows feasible delivery and outcomes consistent with contemporary real-world benchmarks, but pathological response remains limited in biologically unfavorable subgroups. Baseline tumor burden and specific phenotypes identify underperforming populations, supporting prospective validation of risk-adapted perioperative strategies.