Background: Appropriate preclinical evaluation of a bioartificial liver assist device (BAL) demands a large animal model, as presented here, that demonstrates many of the clinical features of acute liver failure and that is suitable for clinical qualitative and quantitative evaluation of the BAL. A lethal canine liver failure model of acute hepatic failure that removes many of the artifacts evidenced in prior canine models is presented.
Methods: Six male hounds, 24-30 kg, under isoflurane anesthesia, were administered 1.5 g/kg D-galactosamine intravenously. Canine supportive care followed a well-defined management protocol that was guided by electrolyte and invasive monitoring consisting of arterial pressure, central venous pressure, extradural intracranial pressure (ICP), pulmonary artery pressure, and end-tidal CO2. The animals were treated until death-equivalent, defined as inability to sustain systolic blood pressure >80 mmHg for 20 minutes despite maximal fluids and 20 microg/kg/min dopamine infusion.
Results: The mean survival time was 43.7+/-4.6 hours (mean+/-SE). All animals showed evidence of progressive liver failure characterized by increasing liver enzymes (aspartate transaminase from 26 to 5977 IU/L; alanine transaminase from 32 to 9740 IU/L), bilirubin (0.25 to 1.30 mg/dl), ammonia (19.8 to 85.3 micromol/L), and coagulopathy (prothrombin time from 8.7 to 46 s). Increased lability and elevations in intracranial pressures were observed. All animals were refractory to maintenance of cerebral perfusion pressure even with only moderately elevated intracranial pressure. Severe neurologic obtundation, seen in 2 of 6 animals, was associated with elevations of ICP above 50 mmHg. Post-mortem liver histology showed evidence of massive hepatic necrosis. Postmortem blood and ascites microbial growth was consistent with possible translocation of intestinal microbes.
Conclusions: The improved lethal canine liver failure model presented here reproduces many of the clinical features of acute liver failure. The model may prove useful for qualitative and quantitative evaluation of BALs.