International Journal of Hematology and Oncology
2025, Vol 35, Num 3 Page(s): 186-192
| Back | Table of Contents | PDF | Mail to Author | |
Real-World Predictors of Pathologic Response to FLOT in Patients with Resectable Gastric Cancer: A Retrospective Analysis
Goksen INANC IMAMOGLU1, Enes YESILBAS1, Galip Can UYAR1
Ankara Etlik City Hospital, Department of Medical Oncology
Keywords: Resectable gastric cancer, FLOT, Pathologic response, Predictors of response, Perioperative therapy
Pathologic response to neoadjuvant FLOT chemotherapy is a key prognostic indicator in gastric cancer, yet reliable predictors remain unclear. This study aimed to identify clinicopathologic and biologic factors associated with response and to develop a preliminary predictive model. A single-center, retrospective cohort of 75 patients with resectable gastric cancer treated with perioperative FLOT between October 2022 and September 2025 was analyzed. Clinicopathologic and laboratory parameters were compared between good responders (Ryan 0–1) and poor responders (Ryan 2–3) using Mann–Whitney U and Chi-square tests. A random-forest classifier incorporating pre-treatment variables was built to explore multivariate interactions and feature importance. The median age was 65 years, and 84% were male. Stage III disease was observed in 81% of patients. Good pathologic response occurred in 45 patients (60%). Poor response correlated significantly with stage III, T3–T4, and high-grade tumors (p= 0.001, 0.003, 0.002, respectively) and higher platelet-to-lymphocyte ratio (p= 0.04). In multivariate analysis, advanced T stage (OR 4.39, p= 0.018), high tumor grade (OR 5.24, p= 0.005), and proximal tumor location (OR 0.30, p= 0.04) independently predicted poor response. The random-forest model achieved an AUC of 0.611 with 65% accuracy. Key predictive features were T stage, N stage, CEA level, and BMI. Tumor depth, histologic grade, and location significantly affect FLOT response. Non-proximal tumors showed more favorable outcomes. The modest machine-learning performance highlights the need to integrate molecular and radiomic markers to refine prediction and personalize perioperative therapy.
Goksen INANC IMAMOGLU1, Enes YESILBAS1, Galip Can UYAR1
Ankara Etlik City Hospital, Department of Medical Oncology
Keywords: Resectable gastric cancer, FLOT, Pathologic response, Predictors of response, Perioperative therapy
Pathologic response to neoadjuvant FLOT chemotherapy is a key prognostic indicator in gastric cancer, yet reliable predictors remain unclear. This study aimed to identify clinicopathologic and biologic factors associated with response and to develop a preliminary predictive model. A single-center, retrospective cohort of 75 patients with resectable gastric cancer treated with perioperative FLOT between October 2022 and September 2025 was analyzed. Clinicopathologic and laboratory parameters were compared between good responders (Ryan 0–1) and poor responders (Ryan 2–3) using Mann–Whitney U and Chi-square tests. A random-forest classifier incorporating pre-treatment variables was built to explore multivariate interactions and feature importance. The median age was 65 years, and 84% were male. Stage III disease was observed in 81% of patients. Good pathologic response occurred in 45 patients (60%). Poor response correlated significantly with stage III, T3–T4, and high-grade tumors (p= 0.001, 0.003, 0.002, respectively) and higher platelet-to-lymphocyte ratio (p= 0.04). In multivariate analysis, advanced T stage (OR 4.39, p= 0.018), high tumor grade (OR 5.24, p= 0.005), and proximal tumor location (OR 0.30, p= 0.04) independently predicted poor response. The random-forest model achieved an AUC of 0.611 with 65% accuracy. Key predictive features were T stage, N stage, CEA level, and BMI. Tumor depth, histologic grade, and location significantly affect FLOT response. Non-proximal tumors showed more favorable outcomes. The modest machine-learning performance highlights the need to integrate molecular and radiomic markers to refine prediction and personalize perioperative therapy.
| Back | Table of Contents | PDF | Mail to Author | |