Can lymphovascular invasion be predicted by preoperative multiphasic dynamic CT in patients with advanced gastric cancer?

ABSTRACT

OBJECTIVES: To determine whether multiphasic dynamic CT can preoperatively predict lymphovascular invasion (LVI) in advanced gastric cancer (AGC). METHODS: 278 patients with AGC who underwent preoperative multiphasic dynamic CT were retrospectively recruited. Tumour CT attenuation difference between non-contrast and arterial (ΔAP), portal (ΔPP) and delayed phase (ΔDP), tumour-spleen attenuation difference in the portal phase (ΔT-S), tumour contrast enhancement ratios (CERs), tumour-to-spleen ratio (TSR) and tumour volumes were obtained. All CT-derived parameters and clinicopathological variables associated with LVI were analysed by univariate analysis, followed by multivariate and receiver operator characteristics (ROC) analysis. Associations between CT predictors for LVI and histopathological characteristics were evaluated by the chi-square test. RESULTS: ΔPP (OR, 1.056; 95% CI 1.032-1.080) and ΔT-S (OR, 1.043; 95% CI 1.020-1.066) are independent predictors for LVI in AGC. ΔPP, ΔT-S and their combination correctly predicted LVI in 74.8% (AUC, 0.775; sensitivity, 88.6%; specificity, 54.1%), 68.7% (AUC, 0.747; sensitivity, 68.3%; specificity, 69.4%) and 71.7% (AUC, 0.800; sensitivity, 67.6%; specificity, 77.8%), respectively. There were significant associations between CT predictors for LVI with tumour histological differentiation and Lauren classification. CONCLUSION: Multiphasic dynamic CT provides a non-invasive method to predict LVI in AGC through quantitative enhancement measurement. KEY POINTS • Lymphovascular invasion rarely can be evaluated preoperatively in advanced gastric cancer (AGC). • Δ PP and Δ T-S were independent predictors for LVI in patients with AGC. • Δ PP and Δ T-S showed acceptable predictive performance for LVI. • Combination of Δ PP and Δ T-S improved predictive performance for LVI. • Multiphasic dynamic CT may be a useful adjunct for detecting LVI preoperatively.