Construction and validation of a nomogram for predicting lymphovascular invasion in advanced gastric cancer: a retrospective study of 1019 patients

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Abstract This study aimed to identify independent risk factors for LVI in advanced gastric cancer (AGC) patients and to develop a nomogram for predicting LVI risk, thereby facilitating individualized clinical risk assessment. Clinicopathological data were retrospectively collected from 1019 AGC patients who underwent radical gastrectomy at the First Affiliated Hospital of Shandong Second Medical University between December 2019 and December 2024. Univariate and multivariate logistic regression analyses were performed to identify independent predictors of LVI, and a nomogram was constructed. Internal validation was conducted using 30% of the samples randomly selected via R software (version 4.4.2). The model's discriminative ability and calibration were evaluated using the area under the receiver operating characteristic curve (AUC), the Hosmer–Lemeshow test, and calibration curves. Clinical utility was assessed using decision curve analysis (DCA). Survival analysis was performed using the Kaplan–Meier method and log-rank test. A total of 1019 AGC patients were included. Univariate analysis revealed significant associations between LVI and CA19-9, CA72-4, tumor differentiation, tumor diameter, perineural invasion, tumor deposit, Lauren classification, pT stage, and pN stage (all P < 0.05). Multivariate logistic regression confirmed that tumor diameter (≥ 4 cm) (OR = 1.836, 95% CI 1.023–3.295, P = 0.042), perineural invasion (OR = 7.628, 95% CI 4.144–14.040, P < 0.001), and pN stage (OR = 7.504, 95% CI 4.758–11.837, P < 0.001) were independent risk factors for LVI. The nomogram constructed using these three factors demonstrated strong discriminative ability, with AUCs of 0.938 (95% CI 0.920–0.957) and 0.914 (95% CI 0.875–0.952) in the training and validation sets, respectively. Calibration curves and the Hosmer–Lemeshow test indicated good model fit ( P > 0.5 in both sets). Decision curve analysis showed that the nomogram provided a net clinical benefit superior to both "treat-all" and "treat-none" strategies within threshold probability ranges of 0.21–0.99 for the training set and 0.30–0.99 for the validation set. Survival analysis of 581 patients with follow-up data revealed that overall survival was significantly worse in the LVI-positive group than in the LVI-negative group (log-rank P = 0.037), with a hazard ratio (HR) of 1.74 (95% CI 1.03–2.94), indicating a near doubling of mortality risk in LVI-positive patients. The nomogram based on tumor diameter, perineural invasion, and pN stage effectively predicts the risk of LVI in AGC patients, demonstrating high accuracy and clinical applicability. The strong association between LVI and poor survival outcomes underscores the importance of this predictive tool in guiding individualized treatment decisions for AGC patients.