Restoration of Bile Duct Injury of Donor Livers During Ex Situ Normothermic Machine Perfusion

Abstract

Background. 

End-ischemic ex situ normothermic machine perfusion (NMP) enables assessment of donor livers prior to transplantation. The objective of this study was to provide support for bile composition as a marker of biliary viability and to investigate whether bile ducts of high-risk human donor livers already undergo repair during NMP.

Methods. 

Forty-two livers that were initially declined for transplantation were included in our NMP clinical trial. After NMP, livers were either secondary declined (n = 17) or accepted for transplantation (n = 25) based on the chemical composition of bile and perfusate samples. Bile duct biopsies were taken before and after NMP and assessed using an established histological injury severity scoring system and a comprehensive immunohistochemical assessment focusing on peribiliary glands (PBGs), vascular damage, and regeneration.

Results. 

Bile ducts of livers that were transplanted after viability testing during NMP showed better preservation of PBGs, (micro)vasculature, and increased cholangiocyte proliferation, compared with declined livers. Biliary bicarbonate, glucose, and pH were confirmed as accurate biomarkers of bile duct vitality. In addition, we found evidence of PBG-based progenitor cell differentiation toward mature cholangiocytes during NMP.

Conclusions. 

Favorable bile chemistry during NMP correlates well with better-preserved biliary microvasculature and PBGs, with a preserved capacity for biliary regeneration. During NMP, biliary tree progenitor cells start to differentiate toward mature cholangiocytes, facilitating restoration of the ischemically damaged surface epithelium.

INTRODUCTION

After liver transplantation was established as a successful treatment for end-stage liver disease, long waiting lists became a major challenge. To meet the high demand, the use of marginal donor organs was explored, and donor criteria were extended over the years.^1,2 Biliary complications such as nonanastomotic strictures (NASs), however, remain the main drawback of the use of extended criteria donor (ECD) livers. NAS are associated with high morbidity rates in ECD liver recipients, often resulting in retransplantation or even death.^3,4

The extent of ischemia-induced damage to the bile duct at the time of liver transplantation has been shown to predict the later development of NAS.⁵ Moreover, bile ducts of livers that suffered from NAS have been characterized by ongoing microvascular-determined biliary hypoxia.⁶ These data suggest that ischemia-induced damage before liver transplantation as well as chronic hypoxia due to substantial loss of the peribiliary vascular plexus (PVP) contribute to the development of NAS.

Nonetheless, the rise of machine perfusion (MP) has sparked optimism in the liver transplantation field.⁷ Perfusing the liver before transplantation has resulted in the following: (1) attenuation of ischemia-induced biliary damage and (2) the ability to safely test bile duct viability.^8-10

Preclinical research has shown that bile duct damage may be assessed based on composition of bile and perfusate samples taken during normothermic machine perfusion (NMP). Among a panel of chemical bile components, pH, bicarbonate, and glucose have shown the strongest correlation with histological bile duct damage and have been identified as highly sensitive markers for damage to the PVP and peribiliary glands (PBGs).¹¹ However, to date, the relation between bile composition and bile duct histology has not been studied in a clinical setting. Moreover, formal evidence that biliary restoration and regeneration of ischemically damaged bile ducts occurs during NMP has never been provided so far.

We recently described our clinical experience with ex situ assessment of initially discarded high-risk donor livers, using combined hypothermic and NMP (dual hypothermic oxygenated liver perfusion [DHOPE]-controlled oxygenated rewarming [COR]-NMP protocol).¹² In the present study, we performed a comprehensive analysis of biliary histology in relation to bile composition and aimed to establish whether biliary regeneration occurs during NMP. We hypothesize that bile and perfusate composition correlate with bile duct integrity and histological viability, and that the restoration of ischemia-induced microscopic bile duct damage already occurs during liver MP.

MATERIALS AND METHODS

Donor Livers and Controls

Forty-two donor livers that were declined for transplantation nationwide and offered for research were selected for DHOPE-COR-NMP in our center. Reasons for organ decline were based on a combination of ECD criteria, including, but not limited to the following: DCD donors aged 60 y and older, ICU stay >7 d, donor body mass index ≥ 30 kg/m², steatosis ≥40%, serum-sodium ≥165 mmol/L, serum AST or ALT ≥3× normal upper limit, or functional donor warm ischemia time ≥30 min. The livers were either accepted (n = 25) or secondarily discarded (n = 17) after MP for liver transplantation. A detailed protocol of this procedure has been published previously.¹¹ All patients provided informed consent in compliance with the Declaration of Helsinki and approval from the Medical Ethical Committee was obtained (METc 2016.281 and METc 2014.077). Biopsies were collected before and after MP, and, in case the liver was accepted for transplantation, after reperfusion in the recipient. The first biopsy was taken after static cold storage, during preparation of the liver on the back-table (before MP). The second biopsy was taken at the end of MP, after disconnecting the liver from the pump, or after reperfusion in the recipient during adjustment of the bile duct for the biliary anastomosis (after MP). All bile duct samples studied here corresponded to the extrahepatic bile duct. As controls, we included biopsies of donor livers that met the regular donor criteria for liver transplantation. These samples are routinely taken in our center after static cold storage (during the back-table procedure) (n = 10) and after reperfusion(n = 10) in the recipient. Selection of these biopsies was ad random and included samples collected between 2015 and 2020 that were physically present in our archives and accompanied by information about donor type. We included a second control group of bile duct samples from explanted livers of recipients that underwent liver transplantation for hepatic diseases. This group was carefully selected by our pathologist who described the specimens as normal large bile ducts (n = 4).

DHOPE-COR-NMP

Consecutive DHOPE, COR, and NMP were carried out as described before.¹² In brief, livers underwent DHOPE at approximately 10 °C for 1 h, a rewarming phase up to 37 °C for 1 h, and at least 2.5 h of NMP. For all procedures, a Liver Assist device (XVIVO, Gothenburg, Sweden) was used. In the current study cohort, both Hemopure (HBOC-201, Hemopure, HBO₂ therapeutics, Souderton, Pennsylvania, USA) and a human red blood cell-based perfusion solution were used as a perfusate solution during NMP. Postoperative results were comparable between both perfusate solutions.¹³ During DHOPE, 1 L of 100% of O₂ was supplied, whereas during COR and NMP oxygen supply was adjusted based on arterial PO₂, PCO₂, and venous saturation. In all cases, the liver started to produce bile at the end of COR or at the start of NMP. During NMP, arterial, venous, and bile samples were collected every 30 min for gas analyses (ABL 90 Flex Plus blood gas meter, Radiometer, Brønhøj, Denmark).

Viability Testing

Viability of the liver and biliary tree was assessed during the first 2.5 h of NMP. Viability criteria are outlined in Table 1. Based on the preset cut-off values and experience with previous cases, a decision to either accept the liver for transplantation or secondarily discard the liver was made after 2.5 h of NMP.¹³ The declined livers were disconnected and discarded after the required biopsies were taken. The viable livers that were accepted for transplantation remained connected to the perfusion device during the recipient hepatectomy. Subsequently, the livers were flushed with 2 L of University of Wisconsin Cold Storage solution prior to placement in an ice-cold University of Wisconsin-filled bowl and transferred to the operating room for immediate implantation into the recipient.

TABLE 1. - Viability criteria to transplant or secondarily discard extended criteria donor livers

Parameter	Accept	Intermediate	Decline
Bile production, mL	≥10	5–10	<5
Perfusate lactate, mmol/L	<1.7	1.7–4.0	>4.0
Perfusate pH	7.35–7.45	7.25–7.35	<7.25
Bile pH	>7.45	7.40–7.45	<7.40
ΔpH	>0.10	0.05–0.10	<0.05
ΔHCO3-, mmol/L	>5.0	3.0–5.0	<3.0
ΔGlucose, mmol/L	<−5.0	−3.0 to −5.0	>−3.0

Values were evaluated within 2.5 h of NMP. The intermediate column represents values that can correspond with a transplantable liver when combined with values from the left column. Δ indicates bile value minus perfusate value. NMP, normothermic machine perfusion.

Adapted from van Leeuwen OB et al, 2019.¹²

Immunohistochemistry and Histomorphologic Analyses

All biopsies for (immuno)histochemistry were fixed in 10% buffered formalin, embedded in paraffin, and cut in 3–4-µm sections. Sections for periodic acid Schiff (PAS), Alcian blue, and hematoxylin and eosin were processed according to the standard protocols. For immunohistochemistry, endogenous peroxidase activity was blocked by a 30-min incubation in 3% hydrogen peroxidase. Antigens were retrieved by applying heat-induced antigen retrieval techniques with citrate pH 6.0 (for anion exchanger 2 [AE2], transcription factor Sox9, and glucose transporter 1 [Glut-1]), Tris/HCl pH 9.0, and Tris/EDTA pH 9.0 (for keratin 19 [K19] and Ki-67, respectively) or by applying Proteinase K (Dako, code S3020; Glostrup, Denmark) (for von Willebrand factor). Sections were incubated for 1 h with primary antibodies (Table S1, SDC, https://links.lww.com/TP/C692). Diaminobenzidine was used as a substrate and sections were counterstained with hematoxylin.

For immunofluorescence, nonspecific protein binding was blocked by 5% bovine serum albumin. Specimens were incubated with primary antibodies, incubated for 1 h with labeled secondary antibodies (Jackson ImmunoResearch, West Grove, PA, USA), and finally counterstained with DAPI for visualization of cell nuclei. Adequate negative controls were included for all immunostainings. To assess injury to the bile duct sections, a histological grading system was employed, as described previously.⁵ All bile duct sections were examined in a blinded fashion. (Immuno)histochemical stainings were scanned by a digital scanner (Hamamatsu, Japan) and processed by ImageScope, ImageJ, and QuPath v0.2.0.¹⁴ The area occupied by PBGs (PBG mass) in bile ducts was evaluated in K19-stained sections and expressed as the percentage of the total analysis area. Sox9+, Ki-67+ (proliferation index), Glut-1+, and AE2+ cells were detected in selected areas by a classifier specifically developed for that stain. Mucus production by PBGs was evaluated in PAS and Alcian blue stains and expressed as the % positive pixels of the total PBG area. Microvascular density (MVD) was calculated as the area occupied by von Willebrand factor+ or CD31+ vessels around PBG clusters. Finally, vessels in the separate PVP layers (inner, middle, and outer) were counted and expressed as number of vessels per microscopic field at ×20 magnification.

Quantitative Real-time PCR

Bile duct biopsies taken before and after MP were snap frozen and stored at −80 °C until RNA isolation. The RNA was extracted using RNeasy mini kit (Qiagen, Germany) according to the manufacturer’s instructions. Additional wash steps with phenol-chloroform-isomylalcohol and chloroform-isomylalcohol were added to the standard protocol. RNA concentrations were measured with a NanoDrop ND-1000 full-spectrum (220–750 nm) spectrophotometer (ThermoFisher Scientific, Wilmington, MA, USA). Single-stranded cDNA was synthesized using Reverse Transcriptase System (Promega, Leiden, The Netherlands) according to the manufacturer’s instructions. cDNA was diluted to 2 ng/μL for real-time quantitative PCR analysis. PCR reactions were performed using Taqman Mastermix (Applied Biosystems, ThermoFisher Scientific, Waltham, MA, USA) in a total volume of 10 μL. The following predesigned primers were purchased from ThermoFisher: K19 (Hs00761767_S1), Nanog (Hs02387400_g1), sodium glucose transport protein 1 (Hs01573793_m1), Glut-1 (Hs00892681_m1), Sox9 (Hs00165814), and AE2 (Hs01586776_m1). Thermal cycling and fluorescence detection were performed on a ViiATM 7 Real-Time PCR system (Applied Biosystems). Expression levels were corrected using K19 as reference gene to select specifically cholangiocytes (ΔCt) and compared with baseline (ΔΔCt). Results are displayed as fold change (2^−ΔΔCt).

Bile Viscosity Measurements

Bile samples were stored in −80 °C after collection. Bile viscosity was measured using a Micro-Ostwald viscometer for manual measurements (SI analytics, Xylem Analytics, Germany). Prior to the measurements, samples were placed in a water bath at 37 °C. The measurements were performed in a 37 °C oven, to ensure constant temperature. The flow time was defined as the time necessary for the meniscus of 2 mL bile to move between the start and stop mark on the viscometer. The capillary of the viscometer had a diameter of 0.43 mm and a constant (K) of 0.01105 mm²/s². The following formula was used to calculate the kinematic viscosity:

$$\begin{array}{c}v(viscosityinm{m}^2/\text{s})\\ =K(0.01105m{m}^2/s^2)\\ \times \left(\begin{array}{c}\text{t}(flowtimeins)\\ -\upsilon (HagenbachCorrection)\end{array}\right).\end{array}$$

Statistics

Continuous data are presented as mean±SD or as median and interquartile range. A 1-way ANOVA test was used for comparison of 4 groups. When a difference was identified by the 1-way ANOVA test, a Bonferroni post hoc test was applied to determine the P-value. The Mann-Whitney U test or Student t test was used for comparison of continuous data among 2 groups. For paired data (eg, before and after MP), the Wilcoxon matched-pairs rank test (nonparametric data) or the paired t test (parametric data) was used. The χ² test or the Fisher Exact test was used for comparison of categorical data among groups. The Spearman nonparametric correlation test was used to determine the relationship between 2 variables. A P-value <0.05 was considered statistically significant. Analyses were performed using Graphpad Prism v9.0 (Graphpad Software, San Diego, CA, USA, www.graphpad.com).

RESULTS

Donor Characteristics and Clinical Outcomes

A total of 42 livers were included in this study of which 25 were transplanted and 17 secondarily discarded. Perfusion parameters after 2.5 h of MP are reported in Table S2, SDC, https://links.lww.com/TP/C692. Baseline donor characteristics were not significantly different between the groups, except for gender and static cold ischemia time (Table S3, SDC, https://links.lww.com/TP/C692). In the transplanted group, a significantly higher percentage of donor livers were derived from male donors, compared with the not-transplanted group (75% versus 59.8%, P = 0.02). Furthermore, the static cold ischemia time was shorter in the livers that were transplanted compared with those that were not selected for transplantation (257 versus 293 min, P < 0.01). All included livers were derived from donation after circulatory death donors. Median follow-up of the patients was 41 mo (range 26–61 mo). After transplantation (n = 25), 1 patient developed bile duct complications that were classified as NAS or recurrence of primary sclerosing cholangitis (PSC) (Tables S4 and S5, SDC, https://links.lww.com/TP/C692).

Bile Duct Histopathology of Transplanted and Not-transplanted Livers

In the transplanted group, samples taken directly after MP and after reperfusion in the recipient showed no differences in histological parameters (Table S6, SDC, https://links.lww.com/TP/C692). Therefore, in the following analyses, data from these samples are presented as one and collectively called after MP. Generally, bile duct specimens showed a decreasing gradient of damage from the lumen toward the outer side of the bile duct wall (Figure 1A). After static cold storage, before MP, minimal remaining mucosa (>75% detached) was found in 75% of the transplanted livers. In contrast, complete loss of the mucosa was observed in all the bile duct specimens in the not-transplanted group (Figure 1A–C and Table S7, SDC, https://links.lww.com/TP/C692). After MP, 16% of the bile ducts in the transplanted group showed <75% loss of the mucosa, whereas there was almost no mucosal lining (<25%) in all the bile duct specimens in the not-transplanted group (Figure 1B–D and Table S7, SDC, https://links.lww.com/TP/C692). Before MP, bile duct specimens of the livers that were eventually transplanted showed less damage to the PVP compared with the bile ducts of livers that were not transplanted (25% versus 64.3% of the samples showing >50% of the vessels damaged; P = 0.02) (Figure 1A–C and Table S7, SDC, https://links.lww.com/TP/C692). After MP, periluminal PBGs appeared less damaged, and therefore better preserved, in the transplanted group compared with the not-transplanted group (28% versus 57.1% of the samples showing >75% loss of the periluminal PBGs; P = 0.02) (Figure 1B–D and Table S7, SDC, https://links.lww.com/TP/C692). An injury gradient was most prominent at the location of the periluminal PBGs. Periluminal PBGs had a better-preserved microvascular density when located >200 μm from the lumen (0.03 ± 0.06 versus 0.18 ± 0.3; P = 0.01). Moreover, the distance from the lumen and length of epithelial lining showed a positive correlation (r = 0.37, P < 0.01) and the length of epithelial lining correlated with MVD (r = 0.59, P < 0.01) (Figure 1E).

FIGURE 1.: Injury assessment of bile ducts that were transplanted and not transplanted. A, Hematoxylin and eosin (H&E) staining of bile duct specimens before machine perfusion (MP) of the transplanted and the not-transplanted group. Fewer vascular lesions were observed in the transplanted group (insets). L, lumen. Scale bars: 100 µm. B, H&E staining of bile duct specimens after MP of the transplanted and the not-transplanted group. Less damage to the periluminal peribiliary glands (PBGs) was observed in the transplanted group (insets). L, lumen. Scale bars: 100 µm. C, Comparison of the percentage bile ducts that received a certain score for each bile duct wall component in the transplanted (T) and not transplanted (NT) group before MP. Shown are the 4 major bile duct wall components: mucosa, vascular lesions, periluminal, and deep PBGs. D, Same comparison for the bile ducts in the T and NT group after MP. E, Immunohistochemistry for endothelial cell marker CD31. Periluminal PBG with epithelial damage and no CD31+ vessels near the lumen (green inset), whereas deeper in the bile duct wall epithelial lining was present and vessels appeared intact (red inset). The graph depicts the %CD31+ vessels directly around the periluminal PBGs within 200 µm and beyond 200 µm from the lumen. *P < 0.05;**P < 0.01. MVD, microvascular density.

PBG Preservation, Activation, and Maturation During MP

PBG mass was significantly higher before MP in bile ducts of the transplanted livers (2.9 ± 2.4) compared with bile ducts of the not-transplanted livers (1.2 ± 1.1; P = 0.01). In accordance with this, the number of progenitor cells, visualized by Sox9, was higher before MP in bile duct specimens of transplanted livers (1269 ± 2863) compared with the not-transplanted livers (258.4 ± 289.9; P = 0.04). Moreover, before and after MP, significantly more proliferating cells were found in the PBGs of bile ducts of the transplanted livers (6.3 ± 6.9) compared with the PBGs in the not-transplanted group (0.9 ± 0.9; P < 0.01 and P = 0.04, respectively) (Figure 2A). Colocalization of Sox9 and proliferation marker Ki-67 confirmed that the progenitor cells within the PBGs were the proliferating cell population (Figure 2B), in line with earlier reports.¹⁵ We next investigated the expression of cholangiocyte transporters, which are the main regulators of cholangiocyte absorption and excretion (Glut-1: glucose and AE2: bicarbonate); these markers also identify mature cholangiocytes. Glut-1 expression was significantly higher in bile ducts of transplanted livers (34.0 ± 21.8) compared with livers that were not transplanted after MP (12.2 ± 10.2; P < 0.01) (Figure 3A and B). In parallel, after MP, the percentage of AE2+ PBG cells was significantly higher in the transplanted group (43.2 ± 12.6) compared with the not-transplanted group (24.9 ± 10.2; P = 0.01). Mature cholangiocytes are generally located near the lumen and can be identified by a columnar shape. Deeper-located cholangiocytes, situated in the deep PBGs, can be recognized by their cuboidal shape (Figure S3A, SDC, https://links.lww.com/TP/C692). Columnar-shaped cholangiocytes were more likely to express the markers for Glut-1 and AE2 than the deeper-located Sox9+ cells (Figure 3C and Figure S3B and S3C, SDC, https://links.lww.com/TP/C692). Glut-1 and AE2 expression, detected by immunohistochemistry, did not show a significant difference between biopsies taken before and after MP (P > 0.05 for all groups) (Figure 3B), indicating that no maturation occurred on a protein level. However, RT-qPCR revealed initiation of cholangiocyte maturation during MP on an RNA level (Figure 3D–E). The Glut-1/Sox9 ratio of cholangiocytes (determined by gene expression) increased significantly when comparing biopsies from before and after MP (0.4 ± 0.5 versus 5.6 + 5.2, respectively; P = 0.02).

FIGURE 2.: Preservation and activation of peribiliary glands (PBGs). A, Immunohistochemistry for keratin 19 (K19) (upper panel), Sox9 (middle panel), and Ki-67 (lower panel) of bile duct specimens of the transplanted (T) and not transplanted (NT) group, before and after machine perfusion. Histograms show quantification (mean and SD) for peribiliary gland (PBG) mass, number of Sox9+ progenitor cells, and %Ki-67+ PBG cells (proliferation index). B, Immunofluorescence of Ki-67 and Sox9 confirming colocalization. *P < 0.05; **P < 0.01; ***P < 0.001.

FIGURE 3.: Preservation of cholangiocyte transporters and peribiliary gland (PBG)-cell maturation. A, Immunohistochemistry for Glut-1 (upper panel) and AE2 (lower panel) of bile duct specimens of the transplanted and not-transplanted group, before and after machine perfusion (MP). Scale bars: 50 µm. B, Histograms show quantification (mean and SD) of %Glut-1+ and %AE2+ PBG cells. T, transplanted; NT, not transplanted. C, Mature Glut-1+ and AE2+ cholangiocytes are generally located near the lumen, whereas deeper located PBGs contain Sox9+ and Nanog+ cholangiocyte populations. L, lumen. D, Graphs show relative gene expression of markers for glucose reabsorption (Glut-1 and sodium glucose transport protein 1 [SGLT-1]), bicarbonate secretion (AE2), progenitor cells (Sox9), and pluripotent cells (Nanog). E, Glut-1/Sox9 ratio in bile duct biopsies taken before and after MP. *P < 0.05;**P < 0.01.

Preservation of the PVP

We next examined the PVP in 3 distinct layers: the outer, middle, and inner layer (Figure 4A). No significant differences were noted between the transplanted and not-transplanted groups regarding the extension of the outer and inner layer, although the extension of the inner layer in both the transplanted (15.2 ± 5.0) and not-transplanted group (13.2 ± 4.2) appeared lower before MP than in donor livers that were accepted for transplantation (22.5 ± 7.8; P < 0.01 and P < 0.01, respectively) (Figure S1, SDC, https://links.lww.com/TP/C692). The middle layer appeared better preserved before MP in the transplanted group (7.2 ± 2.1) compared with the not-transplanted group (5.8 ± 1.7; P = 0.02) (Figure 4B). Integrity of the microvasculature surrounding the PBGs was calculated separately (Figure 4A: dashed line). Microvasculature surrounding PBG clusters (MVD) appeared better preserved after MP in the transplanted group (0.9 ± 0.6) compared with the not-transplanted group (0.5 ± 0.2; P = 0.02) (Figure 4C).

FIGURE 4.: Preservation of the peribiliary vascular plexus (PVP). A, Hematoxylin and eosin (H&E) staining of a bile duct specimen from the transplanted group showing the outer layer (dark gray), the middle layer (light blue), and the inner layer (green) of the PVP. A peribiliary gland (PBG) cluster is lined by a dotted line to indicate the location of the microvessels around PBGs. L, lumen. B, Histograms show quantification of the number of vessels per microscopic field 20× of the outer, middle, and inner layer. T, transplanted; NT, not transplanted. C, von Willebrand factor (vWF) staining of bile duct specimens of the transplanted and the not-transplanted group after machine perfusion (MP) showing the microvascular density (MVD) around PBGs. Histogram shows quantification (mean and SD) of %vWF+ area around PBGs. *P < 0.05.

Mucus Production, Bile pH, and Viscosity

Production of (acid) mucus by PBGs at the start of NMP may lead to changes in bile pH, given that PBGs are connected to the lumen through small tubules that facilitate mucus excretion into the bile flow (Figure 5A and B). To exclude the possibility that bile pH is influenced by PBG mucus production in the MP setting rather than by cholangiocyte absorption and excretion, we stained for PAS (neutral mucus) and Alcian blue (acid mucus). In bile duct specimens of transplanted livers taken before MP, significantly more mucus (neutral: 12.2 ± 6.9 and acid: 2.1 ± 2.9) was produced compared with samples taken after MP (neutral: 4.2 ± 5.0 and acid: 2.1 ± 3.0; P < 0.01 and P < 0.01, respectively). There was no difference between the transplanted and the not-transplanted group in mucus production (Figure 5A). No correlation between bile pH and Alcian blue could be found at the start (r = −0.00; P = 0.99) or at the end (r = −0.20; P = 0.42) of NMP, or any time point (data not shown) (Figure 5C). Next, we explored the possibility that addition of mucus to bile would cause a higher bile viscosity. Interestingly, bile at the start of NMP appeared to have a higher kinematic viscosity than after NMP (P < 0.01) (Figure 5D).

FIGURE 5.: Mucus production by the peribiliary glands (PBGs). A, Histochemistry for periodic acid Schiff (PAS) and Alcian blue (Alc-B) of bile duct specimens before and after machine perfusion (MP). Histograms show quantification (mean and SD) of the PAS+ (neutral mucus) and Alc-B+ (acid mucus) area in PBGs. T, transplanted; NT, not transplanted. B, Cartoon shows how bile secretion starts intrahepatically and is modified along the biliary tree with the addition of mucus by the PBGs residing in the large bile ducts. C, pH before and after normothermic machine perfusion (NMP) plotted against Alc-B, showing no correlation between acid mucus production and bile pH. D, Kinematic bile viscosity of the first bile samples that were collected during NMP (start NMP) and after 2.5 h of NMP (end NMP). **P < 0.01;*** P < 0.001.

Representation of Histological Parameters by Chemical Bile Composition

Of all histological parameters, PBG mass appeared to have the strongest correlation with chemical bile measurements. PBG mass before MP correlated positively with bile pH (r = 0.49; P < 0.01) and bicarbonate (r = 0.52; P < 0.01), and negatively with bile glucose (r = −0.60; P < 0.01) after MP (Figure 6A). In general, transplanted livers secreted bile with a higher pH (7.6 ± 0.08 versus 7.4 ± 0.08; P < 0.01) and bicarbonate (25.1 ± 5.0 versus 16.6 ± 3.8; P < 0.01) levels, and lower glucose (12.3 ± 7.2 versus 18.8 ± 5.6; P < 0.01) levels at the end of MP, as can be expected based on the viability criteria (Table 1). Moreover, the count of periluminal PBGs in each bile duct specimen before MP correlated negatively with bile glucose after MP (r = −0.41; P = 0.02) as well as the extension of the middle PVP layer (r = −0.39; P = 0.02) (Figure 6B).

FIGURE 6.: Correlation between bile components and histological parameters. A, Correlation between bile pH, glucose, and bicarbonate after 2.5 h of normothermic machine perfusion (NMP) and peribiliary gland (PBG) mass before machine perfusion (MP). B, Correlation between bile glucose after MP and the number of periluminal PBGs and the extension of the middle peribiliary vascular plexus layer (PVP) before MP. C, Cartoon explaining the procedure that was used to discriminate between bile ducts that could histologically be defined as severely damaged vs moderately damaged with preservation of restorative function. Livers underwent hypothermic (10 °C) and normothermic (37 °C) MP during which bile and perfusate samples were taken every 30 min. Blood gas analyses were immediately performed, and values were tracked for 2.5 h of NMP. After 2.5 h, the decision was made to either transplant or decline the liver based on the viability criteria. Organs that had met the viability criteria included bile ducts that showed a preserved PVP and microvessels surrounding PBGs, whereas the bile ducts of declined livers showed a significantly less extensive PVP. Moreover, bile ducts of transplanted livers had a substantial number of PBGs and progenitor cells left that could proliferate and mature.

DISCUSSION

In this study, we performed a comprehensive histological evaluation of the large bile ducts of high-risk donor livers that underwent NMP. Livers were divided in 2 groups based on the previously established cut-off values of chemical bile and perfusate components, identifying them as either suitable or unsuitable for transplantation. The current study has shown the following: (1) all extrahepatic bile duct biopsies taken after static cold storage display an increasing injury gradient from the outer surface toward the lumen; (2) viability testing using chemical bile and perfusate composition enables discrimination between livers with histologically severely damaged and less damaged bile ducts. Bile ducts from livers that are transplanted based on the viability criteria have a better-preserved vasculature and a higher epithelial regenerative capacity; (3) of all histological end points, PBG mass has the strongest correlation with bile pH, glucose, and bicarbonate; (4) early signs of biliary tree progenitor cell maturation can be observed during MP; and (5) PBG mucus production and bile viscosity are significantly higher at the start of NMP than at the end of NMP.

PBGs form an interconnected network encircling the lumen of the large bile ducts.^16-18 PBGs connect with the lumen and enable regeneration of the surface epithelium after damage by means of proliferation, migration, and maturation.¹⁵ When viewing the large bile duct in a cross-sectional plane, an epithelial maturation axis is evident from the deep PBGs toward the surface epithelium.^17,19 The Sox9+ progenitor PBG cells have a glycolytic metabolism as opposed to the oxidative metabolism of mature cholangiocytes, indicative of a relative resistance against hypoxic injury.⁶ This progenitor population is located deeper in the bile duct wall and is considered the driver of proliferation and regeneration after substantial epithelial damage.^15,20

In the current study, we observed an injury gradient with substantial loss of mature cholangiocytes located near the lumen and preservation of the deeper located PBGs after static cold storage, in line with the radial maturation axis. The not-transplanted group showed a higher injury score of the PVP directly after static cold storage. This may have aggravated the loss of mature cholangiocytes after the start of oxygenated MP, which is supported by a higher injury score of periluminal PBGs in the not-transplanted group after MP.

Bicarbonate secretion and glucose reabsorption are major functions of mature cholangiocytes and are regulated by Glut-1+/sodium glucose transport protein 1+ and AE2+ cholangiocytes, respectively.²¹ The occurrence of both processes can be detected by the analysis of chemical bile and perfusate composition and are incorporated in the viability criteria. The percentage of Glut-1+ and AE2+ PBG cells reflect the preservation of the periluminal PBGs. Although the surface epithelium was almost completely lost after static cold storage, surviving periluminal PBGs may have played a critical role in bile modification. In fact, applying the viability criteria at the end of MP may have specifically selected livers with intact periluminal PBGs.

For swift and adequate regeneration of the surface epithelium after hypoxic injury, preservation of a sufficient number of PBGs is required.^15,22 The increased PBG area, a higher number of Sox9+ progenitor cells, and a higher percentage of proliferating PBG cells in the transplanted group indicate a better-preserved regenerative function than the not-transplanted livers. Moreover, the bile ducts of transplanted livers showed both well-preserved periluminal PBGs and higher numbers of Sox9+ PBG cells. Because hypoxic injury is more likely to damage mature cholangiocytes, it is no surprise that the more resistant deep PBGs also survived in these ducts. This explains why viability testing can discriminate between livers with bile ducts that have a less damaged versus severely damaged PBG network. Moreover, of all histological end points, PBG mass—a measure including both periluminal and deep PBGs—showed the strongest correlation with bile pH, bicarbonate, and glucose corroborating this notion.

Although no epithelial differentiation during MP could be found on a protein level, maturation of cholangiocytes was evident on a transcriptional level. This suggests that differentiating cells are either preserved or differentiation is initiated during NMP. Differentiation is a process that requires oxygen.⁶ This demonstrates, therefore, that NMP sufficiently oxygenates the biliary tree allowing cholangiocytes to differentiate. The timeframe of oxygenated MP may have been too short to show differentiation on a protein level.

The postoperative outcomes are known for patients who received a donor liver from the transplanted group, whereas these outcomes are of course unknown for the donor livers in the not-transplanted group. NASs are one of the most troublesome complications after liver transplantation and early NASs are associated with ischemic damage during the preservation period and subsequent reperfusion injury during the transplant procedure.²³ Recently, ongoing microvascular hypoxia of the large bile ducts has been identified as a critical mechanism underlying the development of NASs.⁶ The transplanted group showed better preservation of the PVP and microvasculature encircling the PBGs. Based on these results, the transplanted group may have been at a lower risk to develop ongoing microvascular hypoxia—and thus, NAS—after transplantation than the not-transplanted group. Out of 25, a total of 12 patients who received a liver after NMP developed biliary complications, of which only 1 could be attributed to NAS. However, the primary transplantation indication of this patient was PSC, and definitive histology of the explant liver could not distinguish between recurrence of PSC and NAS. The cause of anastomotic strictures and bile leakage is likely to be related to surgical damage during the transplant procedure, rather than ischemic damage of the donor bile ducts during procurement and organ preservation^24,25; therefore, we did not report histological features of these patients separately.

PBG mucus production is a response of PBGs to injury.^26,27 Similar to other organs in the digestive system, biliary mucus protects the surface epithelium against pathogens and forms a physical barrier against (toxic) substances from the inside of the lumen.²⁸ Mucus produced by PBGs includes trefoil factors and several mucins. Trefoil factors are mucin-associated proteins produced by PBGs that are suggested to stimulate biliary epithelial repair after damage.²⁹ In addition, trefoil factors interact with mucins and are known to increase the viscosity of mucin solutions to gel-like properties.³⁰ Our results suggest that PBGs produce an abundant amount of mucus in response to ischemia, which may have been released upon reperfusion during MP. Although it cannot be ruled out that the mucus remained present in the donor biliary tree even after thorough flushing per protocol during organ retrieval. The wash-out of mucus is demonstrated by the increased bile viscosity of the first bile that was produced. This mechanism is in line with ischemia-reperfusion injury of the intestinal crypts. There, mucin granulae are slowly replenished after release resulting in impaired mucus production for several hours.²⁸ This could explain the significant difference in PBG mucus production and bile viscosity before versus after NMP. After the liver transplant procedure, reinstatement of PBG mucus production may occur after a certain period. The sudden increase in mucus production may lead to increased bile viscosity, and progress to sludge and cast formation predisposing transplanted livers to posttransplant cholangiopathies.

When comparing bile duct biopsies directly after NMP and after reperfusion in the transplanted group, no differences were noted for the major histological end points used in this study. We, therefore, combined the biopsies of both time points to 1 group. However, the injury scores for bleeding and inflammation exposed the heterogeneity in the after MP biopsies of the transplanted group. The higher scores in this group can be explained by the portion of biopsies taken after reperfusion.

The biopsies used in this study were taken at the most distal end of the extrahepatic bile duct. Bile composition is a function of hepatocytes and cholangiocytes along the entire biliary tree. Because the cholangiocytes lining the biliary tree are heterogeneous in nature, complete loss of surface epithelium at the most distal end does not signify the same loss in the small ducts. In addition, the distal end of the extrahepatic bile duct may have been manipulated during tube insertion introducing technical artifacts. Bile composition may therefore be poorly represented by this part of the bile duct. Unfortunately, we were not able to study the entire biliary tree, which should be considered a limitation of the study. Another limitation of this study is that we only reported the posttransplant outcomes from 1 group (the transplanted group). Livers from the not-transplanted group were discarded after MP and we could therefore only speculate on posttransplant graft survival in this group based on earlier (non)clinical studies using bile composition or bile duct histology to categorize donor livers.

In conclusion, favorable bile chemistry during NMP correlates well with better-preserved biliary microvasculature and PBGs, with a preserved capacity for biliary regeneration. During NMP, biliary tree progenitor cells start to differentiate toward mature cholangiocytes, facilitating restoration of the ischemically damaged surface epithelium. Efforts to decrease biliary damage of marginal donor livers should focus on thoroughly flushing the arterial biliary vasculature and carefully preparing the extrahepatic bile duct during the procurement and back-table procedures.

ACKNOWLEDGMENTS

The authors are thankful to the UPenn Cell and Developmental Biology Microscopy Core and the Center for Molecular Studies in Digestive and Liver Diseases (grant number: NIH P30DK050306) for the use of microscopes and reagents.