Abstract / Summary
Abstract Background Lymph‑node metastasis represents the dominant metastatic pathway and a critical prognostic factor for gastric cancer. Therefore, accurate pre‑treatment evaluation of regional lymph nodes is clinically essential. Nevertheless, preoperative ¹⁸F‑FDG PET/CT exhibits limited diagnostic performance in routine practice. This study hypothesized that such limitation partially originates from drawbacks of conventional diagnostic criteria, and proposed a modified hierarchical criterion combining metabolic and morphological features to improve PET/CT diagnostic performance for lymph‑node metastasis. Methods Of 173 patients with histologically confirmed gastric adenocarcinoma initially enrolled from three tertiary hospitals, 166 patients with 507 evaluable lymph node groups were included in the final retrospective analysis. Analyses were performed at the lymph‑node group level to account for within‑patient clustering. Two diagnostic criteria were compared: Criterion 1 (conventional, purely SUVmax‑based metabolic criterion) and Criterion 2 (modified hierarchical criterion integrating metabolic and contrast‑enhanced CT morphological features). Diagnostic performance metrics including area under the receiver‑operating‑characteristic curve (AUC), accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated. Performance was further stratified by lymph‑node groups and pathological subtypes. Results Compared with Criterion 1, Criterion 2 yielded significantly improved diagnostic performance for lymph‑node metastasis (AUC: 79.80% vs. 62.70%; accuracy: 84.02% vs. 56.41%; sensitivity: 90.79% vs 47.62%; specificity: 69.59% vs 80.05%; PPV: 83.83% vs 78.30%; NPV: 78.67% vs 49.79%; DeLong test, P < 0.01). Subgroup analyses demonstrated superior performance of Criterion 2 across all Lauren subtypes ( P < 0.05). For WHO pathological subtypes, Criterion 2 achieved higher AUC values relative to Criterion 1; the difference reached marginal statistical significance for tubular adenocarcinoma (TAC, adjusted DeLong test, P = 0.053), whereas statistically significant differences were observed for the remaining subtypes (DeLong test, P < 0.05). Regarding individual lymph‑node group, Criterion 2 showed improved diagnostic efficiency for stations 1, 3A, 4SA, 7, 8, and 9 (DeLong test, P < 0.05). Conclusions The diagnostic performance of ¹⁸F‑FDG PET/CT for regional gastric‑cancer lymph‑node metastasis is strongly dependent on the applied diagnostic criteria. The proposed modified hierarchical criterion integrating metabolic and morphological features outperforms the conventional purely metabolic criterion and shows potential for preoperative clinical application.