HPB O04 Segmental perfusion of ex-vivo human liver segments: A 3Rs compliant directly translatable model for pharmacological, bacteriological and genetic research

Background We have previously presented studies describing the development of a human ex-vivo segmental liver perfusion model (AUGIS 2022, Plenary Prize Session). Utilising surgically resected human liver segments which would otherwise be discarded, bacteriological studies examining host-pathogen interactions can be assessed in a controlled experimental setting. The aim of this study was to compare in a human model the conclusions of work published in the Journal of Experimental Medicine which suggested that in a murine model high-virulence serotypes of Streptococcus pneumoniae exhibited an extraordinary ability to evade capture by Kupffer cells (An et al., 2022). Methods Suitable patients undergoing elective liver resections at a single hepato-pancreato-biliary unit were recruited to the TIMOLD trial. Following hemi-hepatectomies, a healthy liver segment was resected from the specimen and flushed with one litre of heparinised CustodiolTM HTK solution on ice. A segmental branch of the hepatic artery (HA) and portal vein (PV) was identified and cannulated. Ten human liver segments were subjected to normothermic machine perfusion for six hours with OxyglobinTM. During perfusion, six segments were infected with different strains and serotypes of Streptococcus pneumoniae. Perfusate and biopsy specimen were analysed to evaluate bacterial clearance and co-localisation with Kupffer cells. Results Human liver segments exhibited normal histology during perfusion for both control and infected liver segments. Although infected segments had a longer warm and cold ischaemia time than controls (p=0.03 and p=0.01, respectively), there was no difference in PV or HA resistance. Furthermore, there was no difference in lactate clearance (p=0.55), glucose regulation (p=0.87), or potassium clearance (p=0.47). Perfusate culture assays demonstrated no bacterial elimination by the liver. Furthermore, immunofluorescent staining of Kupffer cells and bacteria found no co-localisation of high and low virulence serotypes to hepatic macrophages after 30 minutes (p=0.52) and five hours (p=0.14) of infection. Conclusions Human liver segments did not eliminate pneumococcal strains in a serotype-specific manner or demonstrate preferential co-localisation of bacteria with macrophages based on serotype. These findings highlight issues when extrapolating findings from small animal models as demonstrated with the response of the human liver to invading pathogens. The adoption and implementation of an ex-vivo human liver segmental perfusion model entirely eliminates the need for unethical and unnecessary animal experiments. Using perfused ex-vivo human liver from surgically resected specimens provides a metabolically and physiologically stable environment and is unique in providing results which can be directly translated to the clinical setting.