Correlation of Histological Vessel Characteristics and Diffusion-Weighted Imaging Intravoxel Incoherent Motion-Derived Parameters in Pancreatic Ductal Adenocarcinomas and Pancreatic Neuroendocrine Tumors.

Abstract

To investigate the correlation between intravoxel incoherent motion (IVIM) model-derived parameters and histologically determined microvascularity in pancreatic ductal adenocarcinomas (PDACs) and pancreatic neuroendocrine tumors (PNETs). From April 2013 to May 2014, 42 patients with malignant pancreatic tumors were prospectively examined with diffusion-weighted imaging (DWI) magnetic resonance imaging using a breath-hold variant with 9 b values (0-800 seconds/mm). Intravoxel incoherent motion parameters were extracted from 2 types of volumes of interest (VOIs), one VOI encompassing the total tumor volume (TTV) and another VOI corresponding to the histological regional tumor location (RTV). In 36 PDACs and 6 PNETs, microvessel analysis was performed based on representative tissue sections from the resected tumor tissue, and microvessel density (MVD) as well as microvessel area were calculated. We performed a Pearson-correlation test to study the relationship between IVIM-derived and histology-derived parameters. The DWI-IVIM-derived flowing blood volume fraction f was significantly lower in PDACs compared to PNETs (9.9% ± 5.4% vs 15.5% ± 5.2%; P < 0.0001), the diffusion coefficient D was significantly higher (1.2 ± 0.18 × 10 vs 1.03 ± 0.15 × 10 mm/s; P = 0.001). There was no significant difference in the pseudodiffusion coefficient D* (44.9 ± 52.9 ×10 vs 53.8 ± 51.2 × 10 mm/s). Microvessel density was significantly lower in PDACs (36.8 ± 25.9/mm vs 80.0 ± 26.1/mm; P = 0.0005) compared to PNETs. When derived from the RTV, the flowing blood volume fraction f and MVD of PDACs and PNETs showed excellent correlation (r = 0.85). The correlation using the TTV was moderate (0.64). f (RTV and TTV) and microvessel area showed moderate correlation (r = 0.54/0.47). Intravoxel incoherent motion-derived flowing blood volume fraction f and MVD demonstrate an excellent correlation with histological tumor features in PDAC and PNET. Consequently, IVIM DWI may serve as noninvasive marker of tumor vascularity in different types of pancreatic neoplasms.