Introduction:
Sustaining adequate blood supply to the leg musculature is key to the treatment of peripheral artery disease (PAD). Overexpression of targets such as myocardin related transcription factor (MRTF-A) is known to counteract capillary rarefaction, which would otherwise drive symptom progression and disease burden. Immunofluorescence staining of capillary markers such as platelet endothelial cell adhesion molecule (PECAM-1 / CD31) with subsequent signal counting is a common method to quantify microvascular density. Furthermore, detection of pericytes using markers such as Neural/glial antigen 2 (NG2), in relation to endothelial cells, can be used to assess vessel maturation and stability.
Such quantification is most often performed manually and is therefore highly examiner-dependent, resulting in poor inter-rater reliability and comparability. To circumvent the limitations of these standard approaches we propose an algorithm-based analysis software to standardize quantification and analysis of marker distribution. Through fine-tuned application of automated thresholding and morphological operations, we created a high throughput analysis tool that enables us to detect capillaries and quantify the respective attachment of pericytes in various types of tissues efficiently and reproducibly.
Methods:
Histological sections for validating the algorithm were obtained from murine hearts, skeletal muscles, kidneys and brains, all of which exhibit physiological differences in capillary density and microvascular pericyte coverage. The procedure was then repeated using porcine tissue. After validation, the algorithm was deployed to quantify the effect of AAV-mediated MRTF-A delivery in a porcine model of PAD comprising 18 individual pigs. The algorithm was developed in Python (version 3.11.9) making use of its standard library modules. The implementations of core image processing operations such as adaptive thresholding and morphological transformations are provided by the opencv-python (version 4.10.0.84), a package for python bindings of the computer vision library OpenCV. Further mathematical operations on images were done using NumPy (version 2.1.1) and SciPy (version 1.16.1).
Results:
Validation across the four tissue types confirmed the expected differences in vascular architecture. Capillary density was highest in the myocardium, followed by the kidney, skeletal muscle and brain. Conversely, brain tissue exhibited the highest degree of pericyte coverage. Skeletal muscle and heart followed with comparable levels, whereas kidney tissue – as expected – showed the lowest pericyte coverage among the four.
All three cohorts of pigs – diabetic, hypercholesterolemic and wild type – showed a significantly greater increase in capillary density after treatment compared with untreated PAD animals, while the vessel stability did not decrease, as indicated by pericyte coverage. In total, more than 60 000 capillaries were automatically detected and evaluated for pericyte coverage.
