Discussion
Here we demonstrate for the first time the use of NMP of explanted diseased human livers and the applicability for pharmacokinetic research. We successfully perfused 7 cirrhotic livers and 4 non-cirrhotic livers for a period of 360 min, maintaining liver viability and functionality, as indicated by stable flow, bile production, albumin synthesis, bilirubin excretion, proper histology pre- and post-perfusion and stable ALT and AST values. Additionally, the model showed to be useful to study hepatic first pass, clearance, biliary excretion and drug-drug interaction.
The use of NMP has shown to be of great benefit in the field of organ transplantation (11, 17). Over the past years, NMP has become a widely accepted method to assess viability of the donor liver prior to transplantation (18, 19). Many criteria of hepatocellular and cholangiocellular function have been described (e.g. lactate clearance, perfusate AST, perfusate glucose levels and biliary pH, bicarbonate and glucose levels) to establish liver viability based on perfusion results and post-transplantation outcomes, which demonstrates the robustness of the model in perfusion research (13-17). The explanted cirrhotic livers perfused in this study met most of the criteria for hepatocellular function, except for lactate clearance and portal flow whereas other hemodynamic parameters did not show to be significantly affected. As expected, portal flow was lower in cirrhotic livers compared to non-cirrhotic livers as a result of portal hypertension (20). Cholangiocellular function remained intact during 360 min of NMP as shown by the presence of glucose resorption and bicarbonate excretion.
We previously showed the application of NMP of porcine livers to predict drug-drug interaction on plasma exposure of commonly prescribed statin drugs (21). The degree of drug-drug interactions derived from the NMP model were in good agreement with clinical data showing no effect of ischemia-reperfusion injury on the pharmacokinetics of drugs. Literature is scarce regarding human models that predict drug pharmacokinetics especially in cirrhotic conditions. In the current study we demonstrate the application to study hepatic first pass, clearance and biliary excretion in a perfused explanted human liver model. This approach showed to give comparable results with published in vivo human data. In patients with liver cirrhosis, differences can arise in portal vein concentration after oral dosing due to differences in intestinal absorption and/or portal hypertension compared to individuals without liver cirrhosis. However, in this study we administered the same bolus to the portal vein to study differences in hepatic first pass and clearance between the cirrhotic and non-cirrhotic livers. The non-cirrhotic livers showed to rapidly take up rosuvastatin and digoxin from the perfusate with a hepatic extraction ratio of 0.74 and 0.40 respectively which are close toin vivo reported measures of 0.63 and 0.3 respectively (22, 23). In this study, hepatic first-pass and clearance of rosuvastatin and digoxin showed to be the most affected by cirrhosis. Rane et al. (24) reported that the clearance of hepatically cleared drugs with a high extraction ratio are related to blood flow and thus a major decrease in portal flow as in cirrhosis can dramatically affect the first passage of across the liver (25, 26). In addition, in vivo studies demonstrated a high biliary excretion of rosuvastatin of approximately 76.8% as measured by fecal excretion (27). The ex vivonon-cirrhotic livers showed a biliary excretion of 37% in 120 min, extrapolation of the data resulted in 77% total excretion of rosuvastatin which is in line with in vivo data. Interestingly, digoxin showed a relative high biliary clearance in cirrhotic (51%) and non-cirrhotic ( 17%) livers during 120 min of perfusion. In vivostudies have shown that digoxin is extensively renally eliminated (75%) (28). However, multiple studies demonstrated that digoxin is highly involved in the enterohepatic circulation, thereby decreasing thein vivo fecal excretion of digoxin (29, 30). The two other compounds used in this study, furosemide and metformin, which are mainly renally cleared, showed a low hepatic extraction ratio and minor biliary excretion (≤3%) in both cirrhotic and non-cirrhotic which is in line with human in vivo data which showed that biliary eliminated was limited (31, 32). Interestingly, the percentage of biliary clearance was higher, for all compounds, in the cirrhotic perfused livers. This might be due to an elevated bile flow which has been observed in patients cirrhosis and which is confirmed in our model, resulting in a more efficiency biliary clearance (33).
The effect of drug-drug interactions in cirrhotic and non-cirrhotic livers was subsequently determined by using a cocktail of inhibitors. The non-cirrhotic livers showed an increased AUCR for a drug-drug interaction with rosuvastatin (3.52) where values between 2.48 – 5.38 for rosuvastatin with rifampicin as inhibitor have been observed (34-36). Rosuvastatin is mainly inhibited at the hepatic level, since it is an OATP substrate, and therefore the perfused liver model is showing to properly predict the degree of drug-drug interaction for rosuvastatin. Digoxin, showed a high increase in AUCR upon dosing with inhibitors, which is potentially the result of inhibiting uptake via OATP2 (rifampicin as inhibitor) as well as biliary secretion via Pgp (quinidine as inhibitor). However, this is difficult to compare toin vivo observations since s major part of the drug-drug interaction takes place at the intestinal level, when orally absorbed, thereby affecting the portal vein concentration. Still, the observations from this study showed that we could mimic a drug-drug interaction with digoxin in this perfusion model leading to an increased Cmax and AUCR.
Explanted livers obtained during orthotopic liver transplantation are currently only used for pathological assessment and subsequently discarded. While many preclinical and laboratory animal models try to mimic liver diseases as best as possible, many models fail due to a lack of translation to the human situation. Despite the advantages of utilizing explanted livers, such as specific disease characteristics, maintained hepatic architecture and presence of systemic inflammation and collagen content, this NMP explanted liver model is not widely adopted (or has not been reported) (4). Considering the frequency of orthotopic liver transplantation, and the ease of combining this with the relative simple procedure to prepare the liver for perfusion, this model can be widely applied in a variety of research settings (37). For the field of drug development, use of perfused explanted livers might not only contribute to the reduction in the use of laboratory animals, but more importantly enable better predictions on drug pharmacokineticsin vivo . It will increase our knowledge on hepatic blood flow, hepatic first pass effect and biliary clearance of drugs under normal and cirrhotic conditions. Besides changes in blood flow and collagen content, changes in pharmacokinetics of drugs in patients with liver disease might also be the result of alterations in expression levels of transporter proteins and/or metabolic enzymes. In fact, multiple studies have analyzed liver biopsies from patients with liver disease showing alterations in expression of specific proteins relevant for pharmacokinetics (38-41). For instance, Drozdzik et al., showed an increase in Pgp and multidrug resistance protein 4 (MRP4) and decreases in NTCP, OCT1 and OATP1B1 in patients with severe liver disease (39) . Although these studies already provided some hints towards altered pharmacokinetics and metabolism of drugs in patients with liver diseases, ex vivo perfusion of diseased livers offers a unique opportunity to directly study the effect of altered expression levels of transporter proteins and metabolizing enzymes. In this study we used known drug substrates for different important hepatic uptake and efflux transporters. Gaining insight into pharmacokinetic profiles of OATP1B1/1B2, Pgp, BCRP and OCT-1 model compounds is a first step towards studying transporter functions in diseased liver. This information can be easily extrapolated to other drugs which are substrates for one of the transporters assessed in this study. Additionally, for many drugs, dosing advice is currently incomplete for patients with cirrhosis because of lacking evidence or showing major interindividual differences. Studying drug pharmacokinetics using explanted human livers may substantiate dosing advice for this group of patients (42).
In addition to using cirrhotic liver perfusion to study drug pharmacokinetics, this model provides new possibilities to study specific liver functions under different pathological circumstances for multiple applications, e.g. disease specific biomarkers. Currently, there are several examples of using machine perfusion platforms to study organ reconditioning, pharmacological interventions to prevent ischemia-reperfusion injury, defatting steatotic livers or studying the effect of oncolytics (43-47). Recent studies have adapted machine perfusion to demonstrate the possibility to prolong organ perfusion duration (48-50), with perfusion of human and porcine livers for up to 7 days (18). This might lead to increased knowledge regarding disease pathology and enable the ex vivo treatment of discarded or explanted livers with medication or stem cells (51). Such future interventions would certainly increase the availability of suitable donor organs.
In conclusion, we demonstrated for the first time stable NMP of diseased human livers explanted during liver transplantation and discarded donor livers to study hepatic first pass, clearance ,biliary excretion and drug-drug interactions. This approach provided comparable results to published in vivo human data supporting its applicability as a robust ex vivo drug handling model.
Acknowledgements: We thank Elwin Verheij, René Braakman and Pieter Spigt for their help with the LCMS method development of the drug cocktail.