Work

AI-driven histologic analysis of human Achilles tendinopathy provides a roadmap to unravel pathogenesis

Year: 2025

Type: article

Abstract: OBJECTIVES: Achilles tendinopathy is a common source of pain and dysfunction, yet its pathogenesis remains poorly understood. Research on human tendons is hampered by lack of standardisation in tissue sample validation, making interpretation of results challenging. We sought to develop an automated and operator-independent approach to histologically score human tendons. METHODS: We assembled a cohort of 15 tendinopathic and 10 normal control Achilles tendon samples. We stained longitudinal sections with haematoxylin and eosin and Alcian blue and developed a low temperature epitope-retrieval protocol for immunostaining of blood vessels. Histologic sections were scored by pathologists using the current gold standard Bonar score. Whole sections were then analysed with open-source software (QuPath). Histologic features were automatically quantified across the entire section and summarised in the BonAIr score. Tissue from the same patients was subsequently analysed by quantitative polymerase chain reaction and flow cytometry to validate elements of the BonAIr score. RESULTS: cells were found to localise to areas of low collagen alignment. These results align with the concept of stromal cell dysregulation in tendinopathy. CONCLUSIONS: Automated analysis of whole tendon sections refines conventional histopathologic scoring and predicts cellular and molecular changes found in tendinopathy. The BonAIr score should be further developed for standardised assessment of tendons samples across other anatomical locations and different research centres.

Source: Annals of the Rheumatic Diseases

DOI: 10.1016/j.ard.2025.01.027