Prostate Cancer Microvascular Routes: Exploration and Measurement Strategies

Abstract

Angiogenesis is acknowledged as a pivotal feature in the pathology of human cancer. Despite the absence of universally accepted markers for gauging the comprehensive angiogenic activity in prostate cancer (PCa) that could steer the formulation of focused anti-angiogenic treatments, the scrutiny of diverse facets of tumoral blood vessel development may furnish significant understanding of angiogenic processes. Malignant neoplasms, encompassing PCa, deploy a myriad of strategies to secure an adequate blood supply. These modalities range from sprouting angiogenesis and vasculogenesis to intussusceptive angiogenesis, vascular co-option, the formation of mosaic vessels, vasculogenic mimicry, the conversion of cancer stem-like cells into tumor endothelial cells, and vascular pruning. Here we provide a thorough review of these angiogenic mechanisms as they relate to PCa, discuss their prospective relevance for predictive and prognostic evaluations, and outline the prevailing obstacles in quantitatively evaluating neovascularization via histopathological examinations.

Keywords: biomarkers; fractals; microvessel density; prostate cancer; vasculature.

Figure 1

Tumor angiogenesis predominantly manifests through seven distinct mechanisms. Sprouting angiogenesis serves as the archetypical process in both physiological and pathological angiogenesis. In this mode, new vascular branches emerge from pre-existing blood vessels and ultimately infiltrate the tumor tissue via the migration of tip cells and the proliferation of stem cells. Notably, vessel co-option and vasculogenic mimicry are closely tied to tumor invasion, metastasis, and resistance to traditional anti-angiogenic therapies. Intussusceptive angiogenesis is characterized by the formation of a dual lumen that subsequently bifurcates into two separate vessels, which infiltrate the tumor tissue. Vasculogenesis, on the other hand, entails the recruitment of either bone marrow-derived or vessel wall-resident endothelial progenitor cells. These progenitor cells undergo differentiation into ECs, contributing to the formation of new vascular networks. Vasculogenic mimicry represents another unique avenue, wherein tumor cells extend to create a simulated vascular lumen. These simulated lumens then integrate into pre-existing blood vessels, thereby facilitating the transport of erythrocytes and oxygen into the tumor tissue. Lastly, there exists another process, known as the trans-differentiation of cancer stem-like cells. In this mechanism, these specialized cells acquire endothelial phenotypes and transform into endothelial-like cells via a process known as epithelial–endothelial transformation. Each of these angiogenic modes holds distinct implications for tumor growth, metastatic potential, and responsiveness to anti-angiogenic therapies. As such, understanding these varied processes is crucial for devising more effective treatment strategies and improving prognostic outcomes.

Figure 2

A tissue section of PCa was digitized employing a Zeiss Axioscan.Z1 microscope with a 20× objective lens (A). Vascular structures within the tissue were identified through the use of monoclonal antibodies targeting CD34. A computer-assisted image analysis system was utilized to ascertain the x-y coordinates of each identified vessel, thereby generating a two-dimensional spatial map of their distribution. The system additionally identifies a specific sub-region, delineated by a red outline, where the highest density of vessels is found; this area is the “hot-spot” (B). Within the tissue’s microenvironment, vessels are not uniformly distributed. This leads to a phenomenon known as vascular spatial heterogeneity, where areas rich in vessels are in proximity to areas with fewer or no vessels at all (C). This distribution pattern is influenced by a complex interplay of multiple variables, which are not only interconnected but can also vary over time and space. The irregular configuration of the vascular network poses a significant impact on microscopic examinations. It contributes to three forms of variability: intra-sample, inter-sample, and inter-observer. These variabilities are manifested during the qualitative assessment of tissue slides that have been stained for observation.

Figure 3

In Euclidean geometry, the dimensionality of objects is classified as follows. A point has a dimension of 0, a line has a dimension of 1, a surface is two-dimensional, and a solid object is three-dimensional (A). On the other hand, self-similar fractal objects exhibit non-integer dimensions. For example, the Cantor set has a dimension of 0.63, the Helge von Koch curve features a dimension of 1.2619, the fractal pyramid has a dimension of 2.3219, and the Menger Sponge possesses a dimension of 2.73 (B). Fractal mathematical trees serve as another illustrative example (C). These structures are generated through the continual iteration of a specific equation, resulting in a tree-like figure that exhibits similar characteristics across varying spatial scales. This notion of self-similarity extends to natural phenomena as well, such as the tree ramification, the vascular system in leaves, or two-dimensional vascularity in histological sections (D). These natural systems share key features with fractal objects; they exhibit irregular shapes, possess statistical self-similarity, have non-integer dimensions, and demonstrate the property of scaling. The attribute of scaling implies that the properties of these systems are scale-dependent; the characteristics manifest differently depending on the scale at which they are measured or observed. This underscores the complexity and adaptability of both mathematical and natural fractal systems.

Figure 4

A computational method for assessing the surface fractal dimension of vascular structures in two-dimensional biopsy samples. (A) Sections of PCa tissue are stained with CD34-specific antibodies that selectively bind to blood vessels. (B) Through image segmentation, vessels that display immunoreactivity are isolated based on color similarity among adjacent pixels. (C) The box-counting algorithm is employed to determine the fractal dimension, denoted as Ds. This technique involves quantifying the number of square units, each with side length ε, necessary to fully encompass the target object, represented as N(ε). (D) A representative curve generated via the box-counting method delineates specific fractal windows, determined by the box sizes ε1 and ε2, which serve as the optimal interval for calculating the fractal dimension. When box sizes exceed ε2, they approach the image’s overall dimensions, ultimately resulting in a single box that completely covers the image. At this juncture, N(ε) becomes 1, and the slope of the curve reduces to zero. Conversely, box sizes smaller than ε1 approximate the image’s resolution or a single pixel; in this domain, the box-counting metric essentially reflects the image’s total area.