We have observed epithelial loss from the luminal surface of the bile ducts, often in conjunction with underlying stromal necrosis, in livers donated after circulatory death (DCD) undergoing ex situ normothermic machine perfusion (NMP).1 On occasion, the overlying epithelium has been intact but the underlying stroma necrotic, suggesting a vascular occlusive process as opposed to an intraluminal cause. With Research Ethics Committee approval (14/EE/0137), we perfused 5 DCD livers for at least 4 h using the metra (OrganOx, Oxford, United Kingdom). Livers were then formalin-fixed, and the extrahepatic biliary tree and major septal ducts dissected out and cross-sections examined, noting stromal necrosis on hematoxylin and eosin staining, intravascular fibrin deposition with Martius Scarlet Blue (MSB) stain, and vessel density with CD31 stain. Histological appearances were normal in 1 liver; 4 had stromal necrosis affecting hilar and septal ducts involving between 25% and 100% of the circumference. Fibrin/erythrocyte plugs were visible on MSB in between 25 and 50% of vessels where there was stromal necrosis, but not in its absence (Figure 1A and B are typical examples). In total, 26 duct profiles were reviewed in the 4 affected livers, and 24 exhibited necrosis/fibrin plugs. A second series of 5 liver perfusions were performed with 50 mg tissue plasminogen activator (TPA, Alteplase, Boehringer Ingelheim, Germany) and 250 mL fresh frozen plasma (FFP) infused into the perfusate; no intravascular fibrin and no septal duct necrosis was observed in any liver (Figure 1C and D). Lastly, TPA alone was added to 2 livers with similar appearances to the no-TPA group.FIGURE 1.: Hematoxylin and eosin (H&E) and Martius Scarlet Blue (MSB) stained sections of bile ducts. A, H&E stained section showing widespread stromal necrosis and epithelial loss in an intrahepatic bile duct. Eosinophilic inclusions are present in the subendothelial vessels (arrowed in the magnified panels). B, Corresponding MSB stained section showing that the same subendothelial areas are intravascular aggregates of fibrin and erythrocytes (also arrowed in the panels). These stain red and yellow; collagen stains blue with MSB. C, H&E section of intrahepatic bile duct from a liver treated with fresh frozen plasma (FFP) and tissue plasminogen activator (TPA). The stroma is viable with no subepithelial eosinophilic inclusions. D, Corresponding MSB stained section of the same liver showing no fibrin/erythrocyte intravascular aggregates.Intravascular deposition of fibrin has been noted to occur de novo after NMP in peripheral biopsies of DCD livers and kidneys.2,3 These observations, coupled with observations from our small series of perfusions, need further and more detailed investigation, with clinical evaluation, but have prompted us to suggest the following hypotheses/observations: Bile duct stromal infarcts occur secondary to occlusion of peribiliary vessels with fibrin/erythrocyte plugs. The fibrin/erythrocyte plugs develop after reperfusion as a consequence of ischemia, which is more profound in DCD livers. Stromal infarcts seen to occur during ex situ perfusion may correspond to nonanastomotic biliary strictures (NAS) in transplanted livers. FFP provides an exogenous source of plasminogen enabling TPA to generate sufficient plasmin to clear fibrin deposits in peribiliary vessels. Ex situ treatment of livers undergoing NMP with the combination of FFP and TPA appears to prevent the occurrence of stromal infarcts and may therefore prevent nonanastomotic biliary strictures. NAS have long been thought to be related to bile duct ischemia and occur even where donor heparinization before withdrawal of treatment is routine, suggesting that intravascular thrombosis per se is not a perimortem event. Administration of TPA into the hepatic artery during implantation is reported to reduce NAS, albeit with an increase in intraoperative bleeding due to its widespread fibrinolytic effect.4 This beneficial effect in vivo may be achievable ex situ without the associated intraoperative bleeding since the TPA half-life is very short (5 min) and the liver flushed before implantation.
Read full abstract