List of abbreviations
ALD Alcoholic liver disease
ALT Alanine aminotransferase
AST Aspartate aminotransferase
AUC Area under the curve
AUCR Area under the curve ratio
CIT Cold ischemia time
Cmax Maximum concentration
HA Hepatic artery
HTK Histidine-tryptophan-ketoglutarate
HBV Hepatitis B virus
HCC Hepatocellular carcinoma
MELD Model for end stage liver disease
NAFLD Non-alcoholic fatty liver disease
NMP Normothermic machine perfusion
OLTx Orthotopic liver transplantation
PBC Primary biliary cholangitis
PBPK Physiologically based pharmacokinetic modelling
PV Portal vein
WIT Warm ischemia time
Introduction
Accurate prediction of drug disposition in patients with and without hepatic diseases remains difficult, as appropriate models are lacking. The liver plays an important role in drug handling and impairment or alteration of its function may greatly affect multiple processes. Upon first liver pass, after oral administration, drug bioavailability as well as drug clearance may be altered thereby affecting the drug’s efficacy. Studies in liver cirrhosis have shown that increased bioavailability as well as reduced clearance lead to a higher prevalence of adverse drug reactions and drug-drug interactions which can result in safety issues and ultimately an increased risk for hospital admission (1, 2). Therefore, drug dosing should be tailored according to the varying degree of liver dysfunction among patients with liver diseases. However, with the currently available preclinical and clinical models, it remains difficult to quantify the required tailoring of the dose related to the degree and type of liver dysfunction (3).
Established in vitro and animal models are often used to study the pathology and pharmacological characteristics of drugs of varying diseases. However, translation of these findings to clinical practice remains challenging due to, among others, species differences in transporter expression and the difficulty to mimic dynamic liver processes (4, 5). Novel 3D models like liver-on-a-chip and bile duct-on-a-chip models have gained significant interest as a predictive platform to study liver processes due to the incorporation of haemodynamics (6, 7). Although these organ-on-a-chip models hold much promise, they are still in their infancy owing to the difficulty of mimicking (patho)physiological processes in the liver such as portal and arterial blood flow and biliary excretion (7). Normothermic machine perfusion (NMP) systems using human ex vivo whole organs overcome this problem since hepatic architecture is combined with (near) physiological hemodynamics. Thereby, use of human explanted liver whole organ enables to study hepatobiliary processes as well as liver disease specific pharmacokinetics (8-10).
In this feasibility study we explored the applicability of using explanted human diseased livers for NMP research. Besides standard liver function assessment, we explored liver functionality by studying drug pharmacokinetics; hepatic first-pass, clearance and biliary excretion of four model compounds (rosuvastatin, digoxin, furosemide and metformin). These compounds are known substrates for different important hepatic uptake and efflux transporters. We also studied drug-drug interactions of the model drugs with a cocktail of perpetrator drugs.
Materials and Methods
Human livers
Patients undergoing liver transplantation were included in this study. After providing informed consent, the patients approved the usage of the explanted liver for experimental study. The use of explanted liver tissue was approved by the medical ethical committee of the Leiden University Medical Center (B19.040). Patients with polycystic liver disease, with a transjugular intrahepatic portosystemic shunt (TIPS), or waitlisted for recurrent- orthotopic liver transplantation were excluded from participation. Eleven human livers were included in the study. The underlying disease processes of these livers were primary biliary cholangitis (PBC, n=1), non-alcoholic fatty liver disease (NAFLD, n=2), alcoholic liver disease (ALD, n=3), hepatocellular carcinoma in the context of Hepatitis B viral disease (HBV+HCC, n=2). In addition, three discarded non-cirrhotic livers which were declined for transplantation were included in this study. The reasons for decline were; steatosis (n=2) and a occlusion of right hepatic artery (n=1). Immediately following explantation of the recipient diseased liver, a portal and arterial flush with cold Histidine-tryptophan-ketoglutarate (HTK) (Carnamedica, Warsaw, Poland) preservation solution was performed. The period between explantation (i.e. clamping and transection of the portal and hepatic veins as final step of the hepatectomy) and cold flush of the explanted liver (ex vivo ), is described as the warm ischemia time (WIT). After a clean effluent flush, the liver was transported in cold preservation solution to the Organ Preservation and Regeneration room in the OR complex. Here, under sterile conditions, a back table reconstruction of the right and left hepatic artery and portal vein was performed using surplus donor blood vessels, in order to facilitate cannulation (portal vein- 25Fr cannula, hepatic artery – 12Fr cannula) and connection to the machine perfusion device (LiverAssistTM device, XVIVO, Groningen, the Netherlands). Thereafter, the bile duct was cannulated in order to collect bile fractions during perfusion. Additional flush of the hepatic artery and portal vein was performed on the back table.
Normothermic machine perfusion
All human livers were perfused using the LiverAssistTMdevice. The machine consists of two centrifugal pumps which provide a pulsatile flow to the hepatic artery and a continuous flow to the portal vein (11). The system reservoir was filled with 2L perfusion fluid containing 1:1 ratio of 4x packs of human red blood cells and 4x packs of fresh frozen plasma (Sanquin, Amsterdam, the Netherlands). Insulin, sodium taurocholate, heparin and epoprostenol were provided as continuous infusion at a rate of 10U/h, 1041U/h, 10 mL/h (2% w/v) and 8 µg/h, respectively, in order to maintain liver functioning and to facilitate bile flow. Additionally, nutrients (aminoplasmal 10E (B Braun Melsungen AG, Melsungen, Germany) and cernevit (Baxter BV, Utrecht, the Netherlands) were continuously provided (23mL/hr) to keep the liver metabolically active (Supplemental table 1). Gas delivery to the LiverAssistTM consisted of 95% oxygen and 5% carbon dioxide at 1.5 L/min and the temperature was set at 37°C. The non-cirrhotic livers were perfused with a portal pressure of 11 mmHg and the cirrhotic livers required perfusion at 14 mmHg to generate a sufficient portal flow. Mean arterial pressure was set at 50 mmHg. Upon perfusion, additional boluses of sodium bicarbonate and glucose were given when necessary to maintain physiological pH (~7.35-7.45) and glucose (>5mmol/L) levels. After 360 minutes of perfusion, the livers were submerged in formaldehyde and transported the pathology department and were examined according to the institution’s clinical guidelines dependent on the patient’s underlying pathophysiology.
Drug administration during perfusion
Drug clearance in perfusate and bile of the drug cocktail (rosuvastatin, metformin, furosemide and digoxin) were determined in the absence and presence of drug inhibitors (quinidine, rifampicin, cimetidine and probenecid). The selection of the drug cocktail applied in this study was based since the cocktail is representative for different important hepatic and renal uptake and efflux transporters (12). The dosage applied to the system was based on the typical oral dosage and was corrected for the fraction absorbed in the intestine to the portal vein, fraction of metabolism and circulating volume which is shown in supplemental table 2. After 120 min. of perfusion, a slow bolus for 10 min of the drug cocktail was administered via the portal vein at 1mL/min to mimic oral absorption through the gut. Subsequently, perfusate and bile samples were taken for the following 120 min. Arterial samples were taken at t=120, 122, 124, 126, 128, 130, 135, 140, 150, 160, 170, 180, 210, and 240 min. Additional portal samples were taken during the administration of the drug at t=126 and t=130 min to determine the hepatic extraction. Bile samples were collected in 10 minute fractions from 120 min onwards. After 240 min., first a slow bolus 10 min (1mL/min) of inhibitor mix (quinidine, cimetidine, rifampicin and probenecid) was administered to the liver and after 5 min (at t=245 min.), again, a subsequent slow bolus of the drug cocktail was administered via the portal vein. The same sampling schedule for arterial samples and bile samples was followed. Perfusate and bile samples were immediately stored at ≤-70°C until further processing.
Liver function assessment
Hepatic artery and portal vein flow were recorded from the LiverAssistTM machine. Arterial and venous perfusate samples and bile samples (collected under mineral oil to prevent bile exposure to ambient air (13)) were taken hourly to monitor liver viability (pH, glucose, lactate etc.) using a RapidPoint 500 blood gas analyzer (Siemens, Germany). The total bilirubin, alanine aminotransferase (ALT) and aspartate aminotransferase (AST) concentration in the perfusate samples was measured by reflectance photometry (Reflotron-Plus system, Roche diagnostics, Almere, the Netherlands). Perfusate and bile parameters were compared to defined criteria used in clinical transplantation studies; perfusate ALT <6000 and lactate <2.5 mmol/L after 120 min of perfusion, biliary pH >7.5, Δ glucose perfusate – bile <10 mmol/L, Δ bicarbonate bile – perfusate >0 mmol/L (14-16).
Histological analysis
Pre-perfusion (n=2) and post-perfusion (n=2) biopsies were taken for each liver, fixed in 10% formalin and subsequently embedded in paraffin. Slices of 4µm were cut and stained with hematoxylin & eosin (H&E) or Sirius red for examination using light microscopy.
Bioanalysis
The concentration of the drug cocktail was quantified using LC-MS/MS (Waters, Etten-Leur, the Netherlands). Perfusate and bile sample (10 µL) were deproteinized with 100 µL acetonitrile (ACN) with the addition of 10 of µL the isotopically labelled internal standards (1µg/mL). Thereafter samples were vortexed, centrifuged and supernatant was transferred to 96 well plate and dried under nitrogen. Thereafter, samples were dissolved in 100 µL 10% ACN + 0,1% formic acid and injected in to LC-MS/MS for quantification. Details of the LC-MS/MS conditions used are shown in supplemental table 3.
Chemicals
Rosuvastatin, digoxin, furosemide, quinidine were obtained from Sigma-Aldrich (Zwijndrecht, the Netherlands). Metformin and rifampicin and cimetidine were obtained from Bioconnect (Huissen, the Netherlands). Heparin, sodium taurocholate (Sigma-Aldrich, Zwijndrecht, the Netherlands), insulin (Novo Nordisk, Alphen aan den Rijn, the Netherlands) and epoprostenol (Flolan; GlaxoSmithKline Inc, Mississauga, ON, Canada) were obtained as indicated.
Data analysis and statistics
Data obtained during the perfusion studies was analyzed using Prism version 8 (GraphPad, California, USA). Values for the area under the concentration time curve 0 -120 min (AUC0-tau) were calculated using the linear trapezoidal method. The area under the concentration time curve ratio (AUCR) was determined by dividing the AUC125-245 min (with inhibitors) by the AUC0-120 min (without inhibitors). The hepatic extraction ratio was calculated during the 10 min dosing period as following: concentration entering the liver (portal vein) - concentration leaving the liver / concentration entering the liver. Significance of differences between the cirrhotic and non-cirrhotic livers was tested using the Mann-Whitney U test. Data is presented as median and inter-quartile range (IQR) for non-parametric distributed data. P-value below 0.05 was considered significant.