Dynamic contrast-enhanced ultrasound parametric maps to evaluate intratumoral vascularization

Objectives: The purposes of this study were to assess the reliability of parametric maps from dynamic contrast-enhanced ultrasound (DCE-US) to reflect the heterogeneous distribution of intratumoral vascularization and to predict the tissue features linked to vasculature. This study was designed to compare DCE-US parametric maps with histologic vascularity measurements. Materials and Methods: Dynamic contrast-enhanced ultrasound was performed on 17 melanoma-bearing nude mice after a 0.1-mL bolus injection of SonoVue (Bracco SPA, Milan, Italy). The parametric maps were developed from raw linear data to extract pixelwise 2 semiquantitative parameters related to perfusion and blood volume, namely, area under the curve (AUC) and peak intensity (PI). The mathematical method to fit the time-intensity curve for each pixel was a polynomial model used in clinical routine and patented by the team. Regions of interest (ROIs) were drawn on DCE-US parametric maps for whole tumors and for several local areas of 15 mm² within each tumor (iROI), the latter reflecting the heterogeneity of intratumoral blood volume. As the criterion standard correlation, microvessel densities (MVDs) were determined for both ROI categories. In detail, for all iROI of 15 mm², MVD and maturity were divided separately for vessels of 0 to 10 μm, 10 to 40 μm, and greater than 40 μm in diameter, and the results were correlated with the ultrasound findings. Results: Among the 17 studied mice, a total of 64 iROIs were analyzed. For the whole-tumor ROI set, AUC and PI values significantly correlated with MVD (r_AUC = 0.52 [P = 0.0408] and r_PI = 0.70 [P = 0.0026]). In the case of multiple iROI, a strong linear correlation was observed between the DCE-US parameters and the density of vessels ranging in their diameter from 0 to 10 μm (r_AUC = 0.68 [P < 0.0001]; r_PI = 0.63 [P < 0.0001]), 10 to 40 μm (r_AUC = 0.98 [P = 0.0003]; r_PI = 0.98 [P = 0.0004]), and greater than 40 μm (r_AUC = 0.86 [P = 0.0120]; r_PI = 0.92 [P = 0.0034]), respectively. However, the DCE-US parameter values of perfusion and blood volume were not significantly different according to the diameters (AUC: P = 0.1731; PI: P = 0.2918) and maturity of blood vessels. Conclusions: Parametric maps of DCE-US can be reliably established from raw linear data and reflect the heterogeneous histological measures of vascularization within tumors. In contrast, the values of DCE-US parametric maps (AUC, PI) do not allow deduction of heterogeneous tissue features such as the diameters and maturity of vascular networks.