Early detection of liver fibrosis through hepatic stellate cell activation monitoring using Raman spectroscopy and machine learning

The current methodology used to assess a drug's potential to cause severe liver injuries is rather exhaustive and complex. As a result, such injuries are often overlooked during the drug development process, which makes them one of the leading causes of post-market drug withdrawals and restricted use guidelines of drugs. High-content and continuous screening of liver cells with Raman spectroscopy, a non-destructive and label-free technique, presents a promising approach to identify such pathological conditions early in the drug development process. This research investigates the potential of Raman spectroscopy to identify the progression of liver fibrosis, a disease characterized by the excessive formation of scar tissue resulting in liver functions impairment. The main scar-forming cell type in this pathological process is the hepatic stellate cell (HSC). In healthy conditions, those cells retain 80 % of the retinol present in the human body. However, upon liver injury, those HSCs activate, lose their retinoid content, and synthesize an excess of extracellular matrix leading to scar tissue formation. To evaluate HSC activation by Raman spectroscopy, freshly isolated mouse HSCs were cultured in 2D on fused silica substrates for 2, 5 and 10 days. By recording several Raman maps of those cultures, we demonstrated the disappearance of the characteristic retinol peaks upon activation, enabling discrimination between quiescent HSCs, semi-activated and fully activated HSCs using principal component analysis and a standard K-nearest neighbor classifier. This shows that Raman spectroscopy holds promise in the early detection of liver toxicity during drug development by efficiently identifying HSC activation.