Prediction of neoadjuvant chemotherapy response in locally advanced esophagogastric junction adenocarcinoma using a baseline CT feature-based nomogram model

ZHOU Chuanqinyuan; XU Min; GUO Wenwen; CHEN Tianwu

doi:10.19300/j.2025.L21609

Abstract：Objective To develop a nomogram model based on baseline enhanced computed tomography and clinical features to predict the response of locally advanced adenocarcinoma of the esophagogastric junction(AEG)to neoadjuvant chemotherapy(NAC).Methods Clinical features and enhanced CT images of AEG patients(n=168)confirmed by endoscopic biopsy in two medical centers were retrospectively collected.Patients from Center 1 were randomly divided into a training cohort(n=100)and an internal validation cohort(n=34)in a 7∶3 rate,while patients from Center 2 were used as external validation cohort(n=34).All patients underwent a standardized NAC regimen and pre-and post-treatment contrast-enhanced CT scans.Based on the Response Evaluation Criteria in Solid Tumors criteria,patients in the training set were categorized into a disease control group(71 cases)and a disease progression group(29 cases).Differences in clinical characteristics between the two groups were analyzed using t-test,Mann-Whitney U test,χ2 test,and Fisher's exact test.Features with significant differences were included in a binary logistic regression analysis to identify independent predictive factors,which were then used to construct the nomogram model.The predictive performance of the model was assessed using the area under the receiver operating characteristic curve(AUC),sensitivity,and specificity.Calibration curves evaluated the model's accuracy,while decision curve analysis assessed its clinical net benefit.Results In the training cohort,significant differences were observed in cT stage,gross tumor volume(GTV),and Siewert classification between the disease control and disease progression groups(all P＜0.05).Binary logistic regression analysis identified Siewert classification,cT stage,and GTV as independent predictors of treatment response(all P＜0.05).The nomogram model constructed using these factors achieved good performance with AUC＞0.80 in the training,internal validation,and external validation sets.The training cohort exhibited the highest AUC(0.841)and sensitivity(0.935),but the lowest specificity(0.652).Calibration curves demonstrated a strong agreement between predicted and actual probabilities,and decision curve analysis indicated that the model provided clinical net benefit.Conclusion A nomogram model based on cT stage,GTV,and Siewert classification can effectively predict the response of locally advanced AEG to NAC.