Predictive value of baseline CT imaging features combined with serum biomarkers for neoadjuvant chemotherapy response in adenocarcinoma of the gastroesophageal junction

Lei Li; Fei Luo; Xiansheng Cao; Chao Zhang; Qi Liu; Shanqiang Liu; Libo Yu

doi:10.62347/XLSV6197

Predictive value of baseline CT imaging features combined with serum biomarkers for neoadjuvant chemotherapy response in adenocarcinoma of the gastroesophageal junction

Am J Cancer Res. 2025 Apr 25;15(4):1955-1971. doi: 10.62347/XLSV6197. eCollection 2025.

Authors

Lei Li¹, Fei Luo¹, Xiansheng Cao¹, Chao Zhang¹, Qi Liu¹, Shanqiang Liu², Libo Yu³

Affiliations

¹ Department of Gastrointestinal Surgery, Yantai Affiliated Hospital of Binzhou Medical University Yantai 264100, Shandong, China.
² Medical Department, Yantai Affiliated Hospital of Binzhou Medical University Yantai 264100, Shandong, China.
³ Department of Imaging, Yantaishan Hospital Yantai 264000, Shandong, China.

Abstract

Background: Gastroesophageal junction (GEJ) adenocarcinoma, located at the esophagus-stomach junction, poses significant clinical challenges due to its complex physiological structure. Neoadjuvant chemotherapy (NAC) is standard for tumor downstaging, but response variability necessitates reliable predictive markers. This study evaluates baseline computed tomography (CT) imaging parameters and serum markers as predictors for chemotherapy response in GEJ adenocarcinoma.

Methods: A retrospective study included 304 GEJ adenocarcinoma patients treated with the SOX regimen (S-1 + Oxaliplatin) between January 2020 and December 2024. Patients were categorized based on Tumor Regression Grade (TRG) into effective (TRG 0-1) and poor response (TRG 2-3) groups. Baseline CT characteristics were assessed alongside serum markers, including carcinoembryonic antigen (CEA), alpha-fetoprotein (AFP), carbohydrate antigen 19-9 (CA 19-9), and carbohydrate antigen 72-4 (CA 72-4). Multivariate logistic regression identified independent predictors, and a combined predictive model was developed and validated using an external cohort.

Results: The effective treatment group showed significantly lower serum markers (CEA, AFP, CA 19-9, CA 72-4) and distinct CT parameters, including decreased maximum tumor thickness and area, and lower CT enhancement values. Extramural vascular invasion (EMVI) and tumor surface ulceration were associated with poor response. The combined predictive model demonstrated high accuracy, with an area under the curve (AUC) of 0.813 in the training set and 0.846 in the validation cohort.

Conclusion: Baseline CT characteristics, when combined with serum markers, effectively predict NAC response in GEJ adenocarcinoma.

Keywords: Gastroesophageal junction adenocarcinoma; computed tomography; neoadjuvant chemotherapy; predictive modeling; serum markers; tumor regression grade.