Wesselsbron Virus-Induced Hepatitis in Ewes and Lambs Unraveled Through Machine Learning-Driven Digital Histopathology.

Abstract

Wesselsbron virus (WSLV) disease is an important neglected cause of hepatitis in ruminants with potential for zoonotic transmission, yet its histological lesions have been scarcely studied. We performed a thorough machine learning-driven, pathologist-led digital histopathological assessment of WSLV-induced hepatitis in ewes and lambs infected with clade I (rSA999) ( = 6) or clade II (SAH117) ( = 8) strains and a mock group ( = 6). The analysis was performed on immunohistochemical (IHC) staining for T cells (CD3), B cells (PAX5), histiocytes (Iba1), the WSLV nonstructural protein 1 (NS1), and a double stain for arginase 1 and Ki67 to assess the hepatocyte proliferation index (PI). WSLV-infected animals exhibited significantly higher lymphohistiocytic infiltration and higher hepatocyte PI compared to the mock group. The T cell density was 10 folds higher than the B cell density and was more pronounced in the rSA999 group than in the SAH177 group. Digitally quantified parameters positively correlated with WSLV reverse transcription quantitative PCR (RT-qPCR) results and hepatic injury markers (aspartate transferase [AST], bilirubin, and adenosine deaminase [ADA]), indicating that digital histopathology reliably detects liver damage and disease severity. Among the parameters assessed, the positive correlation between the density of Iba1staining and the WSLV viral load in the liver was the strongest, underscoring the prominent involvement of histiocytes in WSLV-induced hepatitis. This study demonstrates the value of digital histopathological analysis in viral-induced hepatitis using formalin-fixed paraffin-embedded (FFPE) tissue, leveraging whole-slide imaging and deep learning (DL) to objectively characterize key hepatic alterations caused by the viral infection.