Vascular amounts and dispersion of caliber-classified vessels as key parameters to quantitate 3D micro-angioarchitectures in multiple myeloma experimental tumors

Blood vessel micro-angioarchitecture plays a pivotal role in tumor progression, metastatic dissemination and response to therapy. Thus, methods able to quantify microvascular trees and their anomalies may allow a better comprehension of the neovascularization process and evaluation of vascular-targeted therapies in cancer. To this aim, the development of a restricted set of indexes able to describe the arrangement of a microvascular tree is eagerly required. We addressed this goal through 3D analysis of the functional microvascular network in sulfo-biotin-stained human multiple myeloma KMS-11 xenografts in NOD/SCID mice. Using image analysis, we show that amounts, spatial dispersion and spatial relationships of adjacent classes of caliber-filtered microvessels provide a near-linear graphical “fingerprint” of tumor micro-angioarchitecture. Position, slope and axial projections of this graphical outcome reflect biological features and summarize the properties of tumor micro-angioarchitecture. Notably, treatment of KMS-11 xenografts with anti-angiogenic drugs affected position and slope of the specific curves without degrading their near-linear properties. The possibility offered by this procedure to describe and quantify the 3D features of the tumor micro-angioarchitecture paves the way to the analysis of the microvascular tree in human tumor specimens at different stages of tumor progression and after pharmacologic interventions, with possible diagnostic and prognostic implications.

In the second case, we divided the original volume into 4x4x4 cubes and removed all the signal voxels from randomly selected 64-voxel subvolumes. This last approach was aimed to mimic loss of areas of faint signal in order to obtain a further vessel disconnection. The depleted volumes were then intersected with sample-specific, caliber-classified masks to get the new vascular arrangements that were then combined into new series of partially reconstituted trees and analyzed for signal amount and dispersion. Only curves obtained by the first approach were reported in the figure since differences between the two approaches were not appreciable and curves were superimposable.
pag. 9   Table S1. Values of the descriptive parameters obtained from the curves of control

KMS-11 tumors.
Vascular amounts and dispersion data from each point of cumulated vascular arrangements were used for each individual sample to obtain an interpolating line by linear fitting. Fitted lines were characterized in terms of R 2 , slope (ℝ), X/Y coordinates of the first (left-end point) and last (right-end point) points and the X/Y length after projection on the respective axes. N = 4 mice; the two z-stacks belonging to the same animal are highlighted by the same color code as in Figure S1.
pag. 11  Table S2. Values of the descriptive parameters obtained from the curves of Sorafenibtreated KMS-11 tumors.

Vascular amounts and dispersion data from each point of cumulated vascular arrangements
were used for each individual sample to obtain an interpolating line by linear fitting. Fitted lines were characterized in terms of R 2 , slope (ℝ), X/Y coordinates of the first (left-end point) and last (right-end point) points and the X/Y length after projection on the respective axes. N = 4 mice; the two z-stacks belonging to the same animal are highlighted by the same color code as in Figure S1.
pag. 12      mice; the two z-stacks belonging to the same animal are highlighted by the same color code as in Figure S1.
Data are from the analysis of eight 3D vascular samples (z-stacks) obtained from two KMS-11 tumor grafts per group of treatment. For each type of treatment, median vascular amounts and dispersion data from each point of cumulated vascular arrangements were used to obtain an interpolating line by linear fitting. Fitted lines were characterized in terms of R 2 , slope (ℝ), X/Y coordinates of the first (left-end point) and the final (right-end) points and the X/Y length after projection on the respective axes together with their relative IQR inside brackets.