Whole-tumour histogram analysis of pharmacokinetic parameters from dynamic contrast-enhanced MRI in resectable oesophageal squamous cell carcinoma can predict T-stage and regional lymph node metastasis

Abstract

ObjectiveTo identify whether whole-tumour histogram analysis of pharmacokinetic parameters from dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) could predict T-stage and regional lymph node metastasis (LNM) of resectable oesophageal squamous cell carcinoma (SCC). Materials and methodsForty-two consecutive patients with confirmed oesophageal SCC underwent thoracic DCE-MRI. Histogram metrics (median, mean, standard deviation [SD], skewness, kurtosis and entropy) of whole-tumour pharmacokinetic parameters including endothelial transfer constant (Ktrans), reflux rate (Kep) and fractional extravascular extracellular space volume (Ve) were generated by the Omni-Kinetics software. Histogram datasets were interpreted using the Mann-Whitney U test and receiver operating characteristic (ROC) statistical analyses. ResultsThe Mann-Whitney U tests revealed that the median, mean and SD of Ktrans, the SD and entropy of Kep, and the median, mean and entropy of Ve of T1-2 stage oesophageal SCC were lower when compared with T3 stage (all Ps < 0.05); and the ROC analysis showed that the entropy of Ve could reliably distinguish T1-2 stage from T3 stage with an area under ROC (AUC) of 0.773. The Mann-Whitney U tests illustrated that the entropy of Ktrans, and the median, mean, SD and entropy of Kep were higher while the skewness of Kep was lower in tumours with LNM than without LNM (all Ps < 0.05); and the ROC analysis demonstrated that the SD of Kep could best identify tumours with LNM with an AUC of 0.702. ConclusionWhole-tumour histogram analysis of pharmacokinetic parameters of oesophageal SCC on DCE-MRI could be used to predict T-stage and regional LNM.