Transplantation Pathology

Frozen section for steatosis

Because macroscopic appearances are not reliable in assessing the severity and type of steatosis, a frozen section of the donor liver is often obtained in cases where fatty change is suspected clinically. ^, Although frozen section assessments are useful in providing an estimate of overall amount of fat, they can be challenging to assess reproducibly and lack sensitivity in the ‘boundary zones’ between the different grades of steatosis severity. Because of problems with the subjective assessment of steatosis, the use of image analysis has been suggested as providing more objective and reproducible findings ^, but the utility of this alternative approach in clinical practice requires further study. More recently the use of artificial intelligence has also been proposed. ^,

Pathogenesis of liver injury in grafts with macrovesicular steatosis

The pathogenesis of graft dysfunction related to macrovesicular steatosis is multifactorial. Hepatocytes distended with large fat droplets can obstruct the sinusoidal microvasculature, resulting in problems with perfusion of the donor liver at retrieval and subsequent reperfusion in the recipient. ^, Steatotic hepatocytes are significantly more sensitive to oxidative injury and death, causing the release of free lipid from hepatocytes, further compromising sinusoidal blood flow and enhancing lipid peroxidation in sinusoidal endothelial cells.

Other consequences of fatty change on graft function include an increased susceptibility of sinusoidal endothelial cells to cold and warm ischaemic damage, depletion in glycogen content and mitochondrial abnormalities in hepatocytes, a shift to necrosis as the main mechanism for cell death after reperfusion (compared with apoptosis as the predominant pathway in non-fatty livers) and an impaired regenerative capacity following cell loss due to PRI. Complement also plays a role in the development of ischaemia reperfusion injury in steatotic livers.

Histological studies of failed liver allografts with severe fatty change have shown large extracellular fatty aggregates associated with areas of hepatocyte necrosis, haemorrhage and disruption of the sinusoidal framework. The release of fat droplets from necrotic hepatocytes may be associated with the formation of cystic lesions resembling peliosis hepatitis (‘lipopeliosis’) ( Fig. 14.1 ). However, electron microscopic analysis has shown that the free lipid is extrasinusoidal, with only more severe cases resulting in rupture into the sinusoidal space. Fatal diffuse pulmonary fat emboli are described in rare patients.

Donor bile duct biopsies

As discussed later, ischaemic-type biliary (ITBL) complications remain an important cause of graft dysfunction following LT, in some cases leading to graft failure. Recipients of livers donated after cardiac death (DCD) are particularly prone to develop nonanastomotic strictures. A number of studies have found that changes seen in donor bile duct biopsies obtained at the time of transplantation can predict the subsequent development of ITBL. Loss of the epithelial lining is a common finding that is not associated with subsequent biliary complications. Histological features that are predictive for the development of biliary complications include loss of stromal nuclei in the duct wall, detachment/loss of epithelial cells in peribiliary glands, mural haemorrhage and necrosis/other lesions involving arterioles in the peribiliary vascular plexus ( Fig. 14.2 ). One study showed that the severity of mural stromal necrosis and vascular injury was greater in bile duct biopsies obtained after versus before graft reperfusion. The presence of inflammatory cells, particularly CD4+ and CD8+ T cells, in the bile duct wall appears to be associated with a lower risk of biliary complications, suggesting that local adaptive immunological responses may play a role in promoting bile duct epithelial recovery. ^,

Post-reperfusion (‘time zero’) biopsy changes

Time zero post-reperfusion biopsies frequently have morphological abnormalities. Sinusoidal endothelial cells affected by cold ischaemia show swelling, and ultrastructurally, there is vacuolation of the cytoplasm, enlargement of fenestrae and blebs in sinusoidal lumen, with detachment and sloughing into the sinusoid in more severe injury. Other changes are generally mild, tending to be most marked in the centrilobular regions, and include hepatocyte ballooning, necrosis or apoptosis of single hepatocytes, neutrophil polymorph aggregates in sinusoids or around hepatocytes and cholestasis. Neutrophil polymorphs are typically seen around damaged hepatocytes and sometimes have an intracytoplasmic location. Larger areas of confluent necrosis are uncommon but may be the harbinger of evolving severe injury. ^, Similar changes may be seen, generally to a lesser degree, in occasional biopsies obtained from donor livers prior to reperfusion. Given the fact that changes seen in ‘time zero’ biopsies reflect a relatively short period of liver injury, it is not surprising that changes seen at this stage are relatively mild. However, in many instances they appear to represent the earliest stages of more severe damage that can be observed in the first few days or weeks following LT ( Fig. 14.3 ).

A recent study found that time zero biopsy grading could be performed reproducibly, after 45–60 minutes of reperfusion. Mild PRI was characterized by occasional hepatocyte detachment and single or rare clustered neutrophils in the sinusoids. Moderate injury showed clustered neutrophils and some hepatocyte necrosis or apoptosis, and severe injury had zonal (usually centrilobular) hepatocyte necrosis, neutrophilic infiltrates in these areas and clustered neutrophils in more distant sinusoids. The presence of severe PRI was predictive of adverse outcomes in the early post-transplant period, including an increased risk of PNF and of death within the first 90 days.

Histological changes in the early post-transplant period

Hepatocyte ballooning is a common finding in the early post-transplant period. It tends to be most marked in centrilobular areas but in severe cases can be seen throughout the lobule. In most cases this lesion can be attributed to the effects of PRI. An association with high serum transaminase levels during the first 48 hours following transplantation supports this.

If ballooning persists beyond the first 2 weeks post-transplant other possible causes should be considered, particularly when associated with centrilobular necrosis (CLN). ^,

Cholestasis is also a common finding in early post-transplant biopsies. The cholestatic changes are most often centrilobular, but in severe cases of PRI, cholangiolar cholestasis (cholangitis lenta ), similar to that seen in sepsis, may also be present. Cholestasis is not specific for PRI and there are many other possible causes, including rejection, small-for-size syndrome (SFSS), viral infection, sepsis, biliary obstruction and drug toxicity. A syndrome of ‘pure’ cholestasis, sometimes also referred to as ‘functional’ cholestasis, has been reported to occur in the absence of any obvious cause. ^{, ,} Some of these cases probably represent a delayed manifestation of PRI and cholestasis gradually resolves, sometimes over a period of several weeks.

Although fatty change is mainly considered to be a pre-existing donor lesion, there is some evidence to suggest that graft ischaemia and reperfusion injury may lead to the development of steatosis in the early post-transplant period. It may be microvesicular in type.

Primary nonfunction and related disorders

‘P rimary graft dysfunction’ (PDF), ‘initial poor function’ (IPF) (also known as early allograft dysfunction) and ‘primary nonfunction’ (PNF) are related terms used to describe grafts functioning poorly in the immediate post-transplant period. ^, A number of donor and recipient factors have been implicated, but damage related to PRI is likely to be a major factor in most cases. These terms should not be used to describe cases in which there are other peri- or postoperative factors such as vascular occlusion or antibody-mediated rejection (AMR), which can also result in graft dysfunction or failure in the immediate post-transplant period. Clinical features include hyperbilirubinaemia, marked elevation of transaminases (>2000 U/L), prolonged INR with haemostatic problems, hypoglycaemia, hyperkalaemia, metabolic acidosis and renal failure. The diagnosis is usually made in the first 24–48 hours following transplantation.

There are problems with establishing precise diagnostic criteria for these three syndromes, accounting for the wide variation in their reported frequency. IPF can be regarded as a less severe form in which there is potentially reversible graft dysfunction, whereas PNF is a more severe form in which there is graft failure incompatible with its survival. PDF has been suggested as a term to describe all grafts that function poorly in the immediate post-transplant period (IPF and PNF). The reported frequency of PNF ranges from 2% to 23% but recent analysis indicates occurrence in less than 5% of transplants. Histological studies of grafts obtained at retransplantation have shown areas of coagulative hepatocyte necrosis, either centrilobular or panlobular in distribution, ^, suggesting an ischaemic mechanism.

It has been suggested that ‘time zero’ biopsies may be of prognostic value in predicting subsequent poor graft function when certain changes (apart from significant steatosis as already discussed) are present. ^{, ,} However, schemas to grade these lesions and their relationships with graft outcomes have varied. ^, Caution is needed when subcapsular wedge biopsies are taken as a baseline assessment as these may contain areas of zonal necrosis, possibly reflecting the susceptibility of the subcapsular region to ischaemic damage or creasing of the liver due to retraction, with no apparent bearing on subsequent graft function. Recent studies have suggested that the use of alternative methods such as metabolomics and genomic profiling to analyse pre- and postimplantation donor liver biopsies may also be helpful in predicting early allograft dysfunction. ^,

As well as direct liver injury, PRI can be involved in the pathogenesis of other post-transplant complications. Induction of an inflammatory state predisposes to the subsequent development of graft rejection. Prolonged cold ischaemia and PRI play a role in the pathogenesis of ischaemic bile duct injury and other complications occurring later in liver allografts.

Hyperacute rejection

This is largely confined to historical studies when ABO-incompatible grafts were transplanted without pretreatment. Histopathological findings vary according to the severity of the humoral reaction and the time at which tissue samples are obtained. In clinical LT, severe coagulopathy precluded liver biopsy in most cases. In cases where liver biopsy was feasible, a typical sequence of events has been observed. ^, The earliest changes in biopsies obtained 2–6 hours following implantation include the deposition of fibrin, platelets, neutrophils and red blood cells in small vessels and hepatic sinusoids. Endothelial injury leads to widespread neutrophilic exudation, congestion and coagulative hepatocyte necrosis which can be seen in biopsies obtained 1–2 days postimplantation. In severe cases this results in a characteristic picture of massive haemorrhagic necrosis (MHN) throughout the liver ( Fig. 14.4 ). Lack of lymphocyte infiltration or other typical features of TCMR is another characteristic feature.

Immunohistological studies have shown deposition of immunoglobulins (IgG and IgM), complement (C1q, C3, C4) and fibrinogen in vascular and sinusoidal endothelium. ^{, ,} These deposits tend to be most marked during the earlier stages of hyperacute rejection, between 2 hours and 2 days after graft insertion, and rapidly diminish thereafter. Examination of failed allografts reveals more focal deposits of IgM and C1q, mostly confined to arteries. ^, One case of possible hyperacute rejection was associated with strong C4d staining of the endothelium of portal vessels, particularly hepatic arteries, with focal staining of central venules and minimal staining of sinusoids.

Acute antibody-mediated rejection

The presence of preformed DSA has been associated with a higher incidence and greater severity of TCMR. ^{, ,} Importantly there may also be atypical histological features that point to the presence of AMR such as portal/periportal oedema, portal haemorrhage, neutrophil-rich inflammatory infiltration in portal tracts and prominent ductular reaction, producing changes resembling those seen in biliary obstruction ( Fig. 14.5A ). ^{, ,} Careful inspection also shows more diagnostic changes in dilated portal microvessels ^{, ,} including capillaries of the peribiliary plexus as well as the portal venules that extend from the portal vein for a short distance into the lobule (inlet venules). Endothelial cells in these vessels show hypertrophy and cytoplasmic eosinophilia indicating activation, which is accompanied by microvasculitis ( Fig. 14.5B ). ^{, ,} This differs from the endothelial injury seen in TCMR because the inflammatory cells are within the lumen rather than beneath the endothelium and the infiltrating cells are macrophages/monocytes, neutrophils and eosinophils rather than activated lymphocytes. ^, Other changes occurring in acute AMR include portal tract eosinophilia, eosinophilic venulitis (portal and central), prominent central venulitis and dilatation of portal/central veins. ^, A scoring system to indicate the likelihood of acute AMR based on these factors has been described. Arteritis is rarely seen in needle biopsies, but may also be a feature of acute AMR. Injury to tiny vessels forming the vasa vasorum of the biliary tract most likely underlies the later development of biliary strictures.

There are usually changes of coexisting TCMR, ^{, ,} sometimes progressing to chronic (ductopenic) rejection ^{, , ,} if not controlled. ^, Changes described in the liver parenchyma include centrilobular hepatocellular swelling, canalicular cholestasis and sinusoidal neutrophil infiltrates, which can mimic the changes seen in PRI. ^{, ,}

Although most episodes of acute AMR present in the early post-transplant period in the setting of preformed antidonor antibodies, a recent study suggested that some patients may have typical histological features of acute AMR associated with the development of de novo antibodies, up to 2 years post-transplant. Follow-up biopsies obtained from these patients at intervals of up to 4 years after the initial diagnosis of AMR showed persisting features of acute AMR such as portal eosinophilia and portal vein endothelial hypertrophy, as well as the development of periportal fibrosis, suggesting possible progression to chronic AMR.

Immunostaining for the complement component C4d, which is deposited at sites of classical complement pathway activation, can be carried out on routinely processed tissues and is used as a marker of AMR in the renal allograft and more recently in liver. ^{, , ,} However, interpretation of C4d staining in liver is more difficult than in other organs because of nonspecific reactivity and differing staining protocols. ^, Deposition of C4d in liver biopsies from both ABO-incompatible and -compatible grafts with features of AMR has been described ( Fig. 14.5C, D ). ^{, , , , , ,} The C4d deposits persist for at least several days but become more patchy. Positive staining is typically present in a strong linear or finely granular pattern, which may be present on both the luminal and basal surfaces of endothelial cells. Several different patterns have been described in the literature and the significance of these is still incompletely understood. Reviewing these in the context of their own extensive experience, Demetris and colleagues have suggested that typically deposition should be both strong and diffuse (present in >50% of tracts and affecting more than 50% of the circumference of the vessels), most commonly staining the endothelium in the proximal microvasculature, that is, portal veins, portal capillaries, hepatic arteries and periportal sinusoids.

A similar approach for scoring C4d staining on a scale of 0–3 has recently been proposed by the Banff Working Group—a C4d score of 3, which is needed to make a definite diagnosis of acute AMR in a recipient of an ABO-compatible graft, requires the presence of C4d staining in >50% of portal microvascular endothelia (portal veins, capillaries—often extending into portal inlet venules or periportal sinusoids) involving >50% of portal tracts. Care is needed interpreting arteries because of nonspecific staining of the elastic lamina. ^, Portal stromal staining has also been described in ABO-incompatible grafts, possibly reflecting more severe microvascular injury and release of immune complexes from the vessels. Staining of central vein endothelium and sinusoids is less frequently seen. ^{, ,} Sinusoidal C4d expression has been suggested to be the most reliable marker of AMR when immunofluorescence staining is carried out on frozen sections, although this observation has not been confirmed in a more recent study. Diffuse cytoplasmic C4d staining is seen in hepatocytes undergoing necrosis, irrespective of the underlying cause, and therefore is not useful in the diagnosis of AMR.

As discussed earlier, C4d immunoreactivity cannot be interpreted in isolation since it can be seen in native livers with a variety of disorders and also in allografts with a variety of insults including PRI, otherwise typical TCMR, CR, vascular thrombosis, biliary obstruction, recurrent viral hepatitis (hepatitis B and C), recurrent autoimmune disease (AIH and PBC) and in plasma cell-rich rejection (DNAIH). ^,

Chronic antibody-mediated rejection

Chronic AMR has not been fully characterized in the liver allograft but reports are emerging suggesting that the presence of high-titre DSA, either preformed or de novo , is associated with late graft inflammation and fibrosis, particularly in the paediatric population as will be discussed later. ^{, ,} Biopsies from liver recipients with DSA have shown significantly increased interface hepatitis and lobular hepatitis, including plasma cell inflammation, with increased HLA class II expression demonstrated in the inflamed areas.

Fibrosis in putative chronic AMR is characterized by a paucicellular ‘densification’ of portal collagen that has been termed ‘portal collagenization’. This may be associated with obliteration of portal veins (portal venopathy). ^, Subsinusoidal fibrosis is also increased in some studies and is often centrilobular in location. ^{, , ,} Progression to fibrous septa and cirrhosis is described in more severe cases. A scoring system for chronic AMR has recently been proposed which incorporates histological features related to interface hepatitis, lobular inflammation, portal tract collagenization, portal venopathy and sinusoidal fibrosis—a high chronic AMR score in combination with DSA positivity identified individuals at risk for graft loss. ^, Positive immunostaining for C4d in portal microvessels is also helpful in supporting a diagnosis of chronic AMR but tends to be less extensive than that seen in acute AMR.

Acute antibody-mediated rejection

In isolation the histological features described earlier are not specific for AMR and the clinical, histological and laboratory findings must be considered together. Massive haemorrhagic graft necrosis now occurs very infrequently and a clinical diagnosis of hyperacute rejection requires exclusion of other causes of graft failure occurring in the early postoperative period, in particular those associated with the syndrome of PNF. In contrast to PNF, most cases of hyperacute rejection follow an initial period of graft function. Acute graft failure in the early post-transplant period associated with the histological picture of MHN has also been described in the absence of any demonstrable humoral mechanism. Nonhumoral factors which have been implicated in this setting include ischaemia related to hepatic arterial kinking, Gram-negative sepsis, a number of opportunistic viral infections including herpes simplex, herpes zoster, adenovirus and enterovirus, recurrent infection with togavirus-like particles and a single organ Schwartzman reaction. Although idiopathic MHN is rarely seen nowadays, occasional cases with a similar pattern of graft injury are still being reported.

The changes of PRI are similar to those of acute AMR, particularly the presence of centrilobular cholestasis and hepatocyte swelling. Inflammatory cells are more prominent in AMR, particularly innate cells such as macrophages, neutrophils and eosinophils which marginate in veins and small portal microvessels. Staining for C4d is useful to demonstrate deposition in microvessels in AMR. Correlation with the transplant operative details, particularly the ischaemic time and donor factors, can also be helpful.

Acute AMR can mimic changes seen in bile duct obstruction, with a ductular reaction, portal oedema and a neutrophil-rich inflammatory infiltrate. The presence of interstitial haemorrhage, portal oedema, microvasculitis and interstitial rather than periductal neutrophilia should prompt C4d staining and testing for DSA; in this context, portal endothelial or stromal C4d staining favours a diagnosis of AMR.

There is an increasing recognition that acute AMR also stimulates TCMR and overlapping histological features may thus occur. Features which favour AMR as the main diagnosis include portal microvasculitis, portal vein endothelial cell hypertrophy, portal eosinophilia and eosinophilic venulitis, whereas lymphocytic portal inflammation and lymphocytic venulitis are less prominent in AMR. Similarly, TCMR with atypical features such as prominent macrophage, neutrophil and eosinophil infiltrates or which is steroid-unresponsive should raise the possibility of AMR contributing to the graft injury and lead to C4d staining and assessment of DSA.

Chronic antibody-mediated rejection

The four diagnostic criteria suggested for probable chronic AMR (interface and/or perivenular hepatitis, fibrosis, C4d positivity and detection of DSA) are recognized as being relatively stringent and have been proposed to facilitate further studies in this area. The diagnosis is straightforward if all features are present but cases without some criteria are more problematic to classify. When C4d staining is only weak or absent but the other changes are present, a diagnosis of ‘possible chronic AMR’ has been proposed by the Banff Working Group. Terms such as ‘idiopathic post-transplant hepatitis’ (IPTH) or ‘unexplained allograft fibrosis’ have also been proposed for late, nonspecific inflammatory and fibrotic changes and are likely to be manifestations of a low-grade rejection in some cases, particularly those with interface hepatitis, probably having contributions from both cellular and humoral arms of the immune response (discussed fully later). Additionally, DSAs are likely to exacerbate other injuries in the long-term graft as outlined later. Resolving these issues requires more study.

Portal tract lesions in T cell-mediated rejection

The three components of the diagnostic triad are portal inflammation, bile duct damage and venular endothelial inflammation (also known as endothelitis, endotheliitis or endothelialitis). At least two of these three features are required for the diagnosis. Because the inflammatory lesions occurring in TCMR can vary considerably in intensity in different parts of a single biopsy specimen, it is recommended that sections be obtained from a series of levels and that a minimum of five portal tracts be available for examination.

Portal inflammation begins as a lymphocytic infiltrate. By the time that rejection presents clinically there is typically a mixed infiltrate of cells including lymphocytes (mostly T cells), large activated ‘blast’ cells, macrophages, neutrophils and eosinophils ( Fig. 14.6A ). All these cell types are also involved in mediating damage to bile ducts and endothelial cells. Studies carried out in the 1990s suggested that the presence of large numbers of eosinophils indicates a more severe form of rejection, less likely to respond to additional immunosuppressive therapy. ^, It is possible that this association reflects the fact that portal eosinophilia may be a manifestation of concomitant acute AMR (as discussed earlier). Plasma cells occur in some biopsies and increase with severity of rejection. The presence of prominent interface hepatitis with spillover of inflammatory cells into the lobule is also a feature of more severe TCMR. ^, Mast cells may be present in varying numbers. ^, Portal tract granulomas are rarely seen.

The initial damage to bile ducts is probably mediated by lymphocytes, but by the time rejection is clinically evident there is usually a mixed infiltrate, in some cases including a prominent component of neutrophils ( Fig. 14.6B ). Bile ducts are typically cuffed and focally infiltrated by inflammatory cells and may show degenerative changes in the form of cytoplasmic vacuolation, pyknosis and focal disruption of the basement membrane. In cases where portal inflammation is particularly intense, bile ducts can be effaced by inflammatory cells and are difficult to identify in routinely stained sections. Immunostaining for bile duct keratins is useful in demonstrating that bile ducts are still present in this situation. In some cases the presence of large numbers of neutrophils, including lumenal aggregates of pus cells, may mimic changes seen in ascending infective cholangitis. Large numbers of neutrophils have also been identified in samples of bile obtained from patients with TCMR.

Venular inflammatory changes are seen in portal and hepatic vein branches. In early or mild cases there is focal lymphoid attachment to the lumenal surface of endothelial cells. In more advanced or severe cases there is subendothelial infiltration, associated with lifting and sometimes disruption of endothelial cells ( Fig. 14.6C ). Cells associated with endothelial damage are mostly lymphocytes. However, a mixed infiltrate resembling that seen in bile ducts may also be present. In most cases endothelial inflammation affects only a small segment of the vessel. Involvement of the entire circumference of the venule is generally confined to cases with severe rejection. Venular endothelial inflammation has generally been regarded as the most specific feature of liver allograft rejection but it is not invariably present, particularly in cases occurring beyond the early post-transplant period. ^, Furthermore, venular endothelial inflammation can be seen in many other conditions in which there is inflammatory infiltration of portal tracts or the liver parenchyma, including viral hepatitis, PBC, AIH and lymphoproliferative diseases. ^,

Arterial lesions including endothelial inflammation and fibrinoid necrosis have been reported but are rarely seen in needle biopsy specimens ( Fig. 14.6D ). When present, they have been regarded as a sign of severe damage, possibly reflecting concomitant AMR, with an increased likelihood of progression to CR. ^{, ,} Angiographic studies demonstrating attenuation of large and medium-sized arteries suggest that these vessels may also be affected.

Bile ductular reaction is commonly seen in biopsies showing features of TCMR ^, and may in part be a response to other portal tract changes, especially bile duct damage. Other possible causes of a ductular reaction in the early post-transplant period include a delayed effect of PRI, acute AMR ^, and SFSS.

Parenchymal changes in T cell-mediated rejection, including central perivenulitis

Lobular inflammatory lesions comprise a spectrum of changes, principally involving hepatic venules and the surrounding liver parenchyma ^{, , ,} for which the term ‘central perivenulitis’ (CP) is now most widely used. ^, In some cases there may be a more diffuse lobular hepatitis or a predominantly sinusoidal pattern of lymphocytic infiltration. Other terms which have been used to describe these changes include ‘central venulitis’, ‘centrilobular necrosis’, ‘centrilobular necroinflammation’, ‘centrilobular alterations’, ‘centrilobular changes’, ‘hepatitic phase’ of rejection and ‘parenchymal rejection’. ^,

Other parenchymal changes frequently seen in association with TCMR include cholestasis, hepatocyte ballooning, fatty change and focal apoptotic (acidophil) body formation. These lesions also tend to be most marked in perivenular regions and in some cases may be causally related to TCMR but, particularly in the early post-transplant period, much of the parenchymal damage is more likely to be related to PRI than rejection. ^,

The histological features and significance of CP are dependent on when it is seen (early or late) and whether portal changes of TCMR are also present. At one end of the spectrum, usually seen in early post-transplant biopsies, central vein endotheliitis is a prominent feature ( Fig. 14.7A ). Portal tract changes of TCMR are typically also present, usually at least moderate in severity. The diagnosis and grading of rejection in these cases are relatively straightforward.

In other cases, usually occurring significantly later post-transplant, ^, centrilobular necroinflammatory lesions are present with little or no central vein inflammation ( Fig. 14.7B ) and sometimes also with minimal or mild portal inflammatory changes—also referred to as ‘isolated CP’(ICP) or ‘isolated parenchymal rejection’. ^{, , ,} A diagnosis of rejection in such cases is less easily established and other causes of centrilobular damage also need to be considered ( Table 14.5 ) ^, —these are discussed in more detail later. In two studies, features of ICP were observed in 22% of children biopsied >3 months post-transplant and 28% of adults undergoing protocol biopsy >3 years post-transplant.

The inflammatory infiltrate is mainly mononuclear with lymphocytes typically predominating, sometimes with conspicuous plasma cells. In the latter case, the rejection variant ‘plasma cell-rich rejection’ is appropriate if the plasma cell component exceeds 30%. Foci of mild congestion and pigmented macrophages can occur. Perivenular necroinflammatory lesions may also be associated with the gradual development of parenchymal fibrosis in zone 3, with linkage in more severe cases. ^,

The prognostic significance of centrilobular inflammation and dropout is dependent on the histological context. There is evidence to suggest that the presence of CP with portal features of TCMR in the first few months after transplantation indicates a more severe form of rejection, which is less likely to respond to IS and is more likely to progress to CR. ^{, ,} In many cases, the centrilobular changes appear to be present at an early stage, before bile duct loss is evident; recognition of this process and instigation of appropriate immunosuppressive therapy may prevent progression to irreversible changes of CR.

On the other hand, when present as ICP the prognosis is generally favourable. Most cases, if treated at all, respond to bolstered IS ^, but the lesion may persist, recur or fluctuate. If there is no response to increased baseline IS or a single steroid bolus, the inflammation may behave more like plasma cell-rich rejection (previously called de novo AIH) and respond to the reintroduction of steroids or mycophenolate for a period of time rather than more aggressive depleting therapies such as OKT3. When untreated, there may be progression to centrilobular fibrosis. ^, One recent study suggested that patients with ICP (‘isolated parenchymal rejection’) had an increased risk of developing CR. A grading scheme for the severity of CP proposed by the Banff Working Group has been shown to correlate with adverse outcomes in one study of ICP.

In some cases inflammation of hepatic venular endothelium may be associated with the development of veno-occlusive lesions and more severe congestive changes resembling venous outflow obstruction ( Fig. 14.8 ). ^{, ,} Many of these cases have TCMR with particularly severe endothelial inflammation but the lesion can also develop more insidiously, without overt portal changes of TCMR, although it still appears to be responsive to optimization of IS.

Late T cell-mediated rejection

Several studies have suggested that late TCMR may have different histological features from early or ‘classical’ TCMR. These include a predominantly mononuclear portal inflammatory infiltrate (contrasting with the mixed population of cells seen in early biopsies), less inflammation of bile ducts and portal venules and more prominent interface hepatitis. ^{, , ,} The overall appearances may thus come to resemble those seen in chronic viral hepatitis or AIH. As discussed earlier, centrilobular necroinflammatory lesions falling within the spectrum of CP, either with or without portal inflammation, are a common feature of late TCMR.

Cases with predominantly lobular changes often present with raised transaminase levels, contrasting with the cholestatic profile that is more typically seen in early portal-based acute rejection, or the LFTs may be normal. ^, However, the presence of cholestasis may identify patients with a more severe form of late TCMR, which is less likely to steroid responsive. Late TCMR is associated in many cases with inadequate IS ^{, , ,} and there is an associated increased risk of developing a number of adverse outcomes including further episodes of TCMR and progression to CR. ^{, ,} Those with prominent interface hepatitis and/or CP can progress to the picture of plasma cell-rich rejection (DNAIH) or develop gradual centrilobular fibrosis. ^{, , ,}

Histologically there are similarities between late TCMR, plasma cell-rich rejection, chronic AMR and IPTH, suggesting that these conditions may all be part of an overlapping spectrum of late immune-mediated damage in the liver allograft.

Portal tract changes

Two main diagnostic features have been described. These are (1) dystrophic biliary epithelial changes or loss of bile ducts from greater than 50% of portal tracts and (2) foam cell arteriopathy. ^{, ,} Some cases of late CR may have different histological features, including CH-like changes. ^{, ,}

During the early stages of CR the bile ducts show inflammatory infiltration, indistinguishable from that seen during potentially reversible TCMR. In cases where there has been an incomplete biochemical response to additional IS, there may be a reduction in the overall degree of inflammation in portal tracts. However, bile ducts show continued lymphocytic infiltration associated with nuclear pleomorphism, uneven nuclear spacing, disordered polarity and focal attenuation and/or disruption of biliary epithelium. The cells are enlarged with eosinophilic cytoplasm, and with the enlarged, irregular nuclei this produces a dystrophic or atrophic appearance ( Fig. 14.10A ). These changes are associated with features of replicative senescence (e.g. nuclear p21 expression) and are widely regarded as an early sign of impending bile duct loss. ^, When these appearances are seen without loss of the majority of bile ducts, there is a greater likelihood of reversal. Interestingly, one study suggested that the expression on biliary epithelial cells of the senescence-associated factor p21 and a marker of mesenchymal differentiation (S100A4) may be occurring as a relatively early stress-related response in liver allograft rejection, which is potentially reversible if the hostile environment is modulated.

As the disease progresses to a later stage, there is loss of bile ducts, typically associated with a diminishing cellular infiltrate that eventually produces a characteristic ‘burnt out’ appearance in end-stage livers ( Fig. 14.10B ). Bile duct loss principally affects the small interlobular bile ducts and is thus readily diagnosed in needle biopsy specimens. In normal liver allografts, at least 70–80% of portal tracts should contain a bile duct of equivalent diameter to the hepatic artery branch. ^, Bile duct loss should occur in more than 50% of portal tracts to make a firm diagnosis of late CR. However, there are problems in counting bile ducts accurately, particularly in small needle biopsy specimens. Loss of both the artery and bile duct from a portal tract may also impede counting. ^{, ,} An adequate sample, usually 16G passes with at least 11 portal tracts and 20–30 mm combined length and ideally containing 20 or more portal tracts, and/or the demonstration of ductopenia in several biopsies may be required before a confident diagnosis of bile duct loss can be made. In hepatectomy specimens obtained at retransplantation there may be loss of epithelium from medium-sized (septal) and large (hilar) ducts. The latter sometimes show a distinctive pattern of lumenal obliteration by fibrous tissue and inflammatory cells ( Fig. 14.10C ).

A notable feature is the absence of bile ductular reaction or periportal fibrous expansion in most cases, the main exception being when a distal biliary stricture is associated. This contrasts with other diseases associated with loss of bile ducts, in which these secondary changes are nearly always present. Studies have attributed the lack of ductular reaction in CR to an increase in apoptosis or a reduction in proliferative activity ^, within the ductular compartment. A close relationship has been observed between ductular reaction and periportal neovessel formation in a number of liver diseases and a lack of these two reparative responses has been implicated in the pathogenesis of irreversible bile duct loss in CR. Staining for biliary keratins such as keratin 7 helps to confirm the absence of bile ducts and lack of a ductular reaction in CR and often also shows prominent positive staining of periportal cells with an intermediate hepatobiliary phenotype ( Fig. 14.10D ). This immunostaining also highlights the frequent loss of periportal canals of Hering, the most terminal branches of the biliary system.

The characteristic vascular lesions of CR are seen in large and medium-sized arteries and are typically manifest as intimal aggregates of lipid-laden foamy macrophages ( Fig. 14.11A ), although other layers of the arterial wall can also be affected. These occlusive arterial foam cell lesions may produce abnormalities that can be detected angiographically. In some cases with a more acute presentation, there is a prominent infiltrate of inflammatory cells ( Fig. 14.11B ), mainly T lymphocytes, suggesting an overlap with TCMR. Conversely, in cases with a more prolonged course, there are increasing numbers of myofibroblasts associated with varying degrees of intimal fibrosis as well as fragmentation of the internal elastic lamina. However, advanced fibromuscular intimal thickening of the type classically seen in end-stage CR affecting renal or cardiac allografts is only rarely found in liver allograft rejection ( Fig. 14.11C ). The macrophages and mesenchymal cells in arterial lesions are of recipient origin. Because these arterial lesions rarely affect small vessels of the size sampled in needle biopsy specimens, the definitive diagnosis of chronic vascular rejection is usually only made when the whole liver is available for examination ( Fig. 14.11D ). However, smaller portal tracts may show a reduced number of small arterial branches and other microvascular channels. ^{, ,} These changes can occur during the early stages of CR, before bile duct loss is present.

Very rare cases show only isolated vascular lesions, either inflammatory or fibrointimal thickening. These were found to be a marker of imminent TCMR or chronic AMR and warrant close follow-up.

Inflammatory and/or foam cell lesions are also seen in portal and hepatic venules in some cases of CR, particularly those associated with a more acute presentation, again suggesting that there are areas of overlap with TCMR. ^, Inflammatory lesions in hepatic venules may also result in fibrous lumenal obliteration, producing changes similar to those seen in hepatic veno-occlusive disease (VOD). ^{, ,} Similar changes can also occur in small portal vein branches ( Fig. 14.12 ). A combination of fibroinflammatory occlusive lesions involving portal and hepatic veins appears to be important in the pathogenesis of the parenchymal fibrosis in CR, in some cases resulting in a cirrhosis-like appearance, with a venocentric pattern. ^, A similar mechanism has been postulated for the development of fibrosis and cirrhosis in nontransplanted liver.

In most cases of CR, bile duct loss and occlusive arteriopathy are both present. Morphometric studies have demonstrated a parallelism between the severities of these two components of CR and have suggested that ischaemia may be a factor contributing to bile duct loss in the liver allograft. However, there are well-documented cases with a purely ductopenic or a purely vascular form of CR. In a combined series of 72 cases from three centres, ^{, ,} 51 (71%) had both lesions, 10 (14%) had ductopenia alone and 11 (15%) had a purely vascular form of CR.

Parenchymal changes

Perivenular bilirubinostasis is a prominent finding in CR and, in most cases, is presumably related to bile duct loss. Cholestasis can also be seen with the purely vascular form of CR, suggesting that ischaemia may also be a factor in some cases. Sinusoidal foam cells are commonly seen and are probably also a response to cholestasis.

Perivenular necrosis is also a common finding ( Fig. 14.13A ) and typically occurs as a sequela to the necroinflammatory lesions, which are seen during the preceding phase of TCMR. Humoral mechanisms may also be involved; the deposition of C4d has been described in portal and central venules as well as perivenular sinusoids. ^, In cases where there is incomplete biochemical response to additional IS, the cellular infiltration in perivenular regions often subsides, but hepatocellular dropout persists. Necrosis typically has a lytic pattern and is accompanied by reticulin collapse and immature collagen fibre deposition. In some cases there may be more extensive necrosis with bridging and nodule formation ( Fig. 14.13B ). Even in cases where zonal necrosis has been detected in serial biopsies over a period of several months, the lesions frequently retain an appearance suggesting acute damage, without the formation of mature collagen or elastic fibres. This suggests a dynamic equilibrium between hepatocyte loss and regeneration in perivenular regions. However, in some cases there is development of more mature fibrous lesions, which may ultimately progress to cirrhosis-like changes. ^{, ,} A range of rejection-related ischaemic mechanisms have also been implicated in the pathogenesis of parenchymal necrosis and fibrosis; these include occlusive lesions in large and medium-sized arteries, loss of small arterial branches and occlusive lesions involving portal and/or hepatic veins. Although CLN has been regarded as a ‘surrogate marker’ of rejection-related arteriopathy, a definite association between these two processes has not been convincingly demonstrated.