Chronic Rejection. A general overview of histopathology and pathophysiology with emphasis on liver, heart and intestinal allografts

Introduction

Recent advances in immunosuppression, operative techniques and patient management have improved short term one year survival rates after kidney, liver and heart transplantation to more than 80% at most experienced centers. In fact, organ allograft recipients without major post-operative complications are being discharged from the hospital within several weeks of the operation, a scenario almost unheard of even as little as a decade ago. These developments have shifted considerable attention in research away from short term problems such as acute rejection, to the major obstacles to long-term, morbidity-free survival: chronic rejection (CR), the subject of this review, and the adverse effects of maintenance immunosuppression, which is needed to prevent CR.

The nature and magnitude of the problem

CR can be broadly defined as a largely indolent, but progressive form of allograft injury characterized primarily by fibrointimal hyperplasia of arteries, or obliterative arteriopathy (OA), interstitial fibrosis and atrophy of parenchymal elements. For the most part, it is irreversible and eventually results in allograft failure. Currently, CR is the most significant medical obstacle to long term morbidity-free allograft survival. The incidence is thought to progressively increase with time after transplantation, but after a period of five years it affects about 10% of liver to around 60% of lung allograft recipients (table 1) [1–11].

Table 1.

Estimates of the incidence of chronic rejection in various solid organ allografts

Allograft	Incidence of OA at 5 Years	References
Heart	25% – 60%	[1], [2], [3]
Heart-Lung	8% – 20% (OA)	[1], [4]
	50% – 60% (OB)	[1], [5], [7], [8]
Lung	28/45% (single/double lung)	[7]
Liver	3–17%; 10% and decreasing	[9], [10]
Pancreas alone	30–70%*(late graft loss)	[11]
Pancreas and kidney	20–40%*	[11]

^*Estimates based on graft survival rates after one year, with the assumption that most late graft failures are due to CR.

The term “chronic” implies a temporally prolonged course and in general, CR more indolently compromises organ function, in contrast to acute rejection that can precipitously cause allograft failure because of vascular necrosis/thrombosis and infarction. However, many cases of CR clearly evolve from severe or inadequately controlled acute rejection episodes and in patients not compliant with chronic maintenance immunosuppression. In such patients, persistent immunologic injury leads to a progressive decline in organ function over a period of weeks to months. There is however, a significant number of patients who do not fit these profiles, and slowly develop graft failure over a period of years. This more indolent presentation may be attributable to “clinically silent” rejection episodes that go undetected or to other factors.

Alloimmunity was the suspected etiology of CR when it was first reported over 3–4 decades ago [12, 13]. Indeed, from an immunological perspective, CR could be thought as an iatrogenically-induced autoimmune disease that shares many features with chronic graft versus host disease (CGVHD) However, CR appears to more selectively damage the arterial tree in comparison to autoimmune diseases and GVHD. In fact, obliteration of the arterial lumen because of fibrointimal hyperplasia, or obliterative arteriopathy (OA), is the pathognomonic lesion of CR.

The premise that CR is immunologically mediated, is best supported by the fact that isografts rarely suffer this complication and if they do, it is much delayed and less severe in comparison to allografts [14–17]. The importance of immunologic injury is also supported by many clinical studies, in which severe or persistent acute rejection [10, 18–26], inadequate immunosuppression [8, 27], donor/recipient racial or ethnic differences[28–32], sex mismatching[24, 33], viral infections [34–41] and the use of immune activating drugs are cited as risk factors [42].

The effect of MHC matching on long term allograft survival is also used as evidence of an immune etiology [28, 29,31,32,43–46]. However, the actual impact of imperfect matching on long term allograft survival has been less than expected [28,31,43,44,47–51] With complete six antigen matches, immunological causes of allograft failure are minimized [52,53] Any degree of mismatching usually results in a progressive attrition of graft function over time [28,29,31,32,43–46], which is in large part, related to CR. Liver allografts however, are an exception. They are relatively resistant to CR (table 1) [51] and HLA matching appears to play a dualistic role [54,55]: Mismatching increases the incidence of acute rejection, but MHC matching might contribute to a greater susceptibility to recurrent inflammatory liver disease [54].

Despite a convincing body of data, a level of uncertainty still exists about OA pathogenesis, largely because it is difficult to study in clinical material. Prior to graft failure or death when the entire organ can be histopathologically examined, invasive diagnostic procedures are often needed to make the diagnosis of CR. These tests are often subjectively interpreted. The issue is further confused by atherosclerotic donor disease [56–58]; the development of atherosclerosis after transplantation, and the observation in some studies that OA is associated with conventional risk factors for atherosclerosis [59–62]. Lastly, the various organs are affected differently by CR, so there is no generic definition that can equally applied across all organs. For example, in heart allografts, the primary manifestation of CR is OA and progressive ischemic damage [1,63–65]. In lung allografts, destruction of the small bronchioles (obliterative bronchiolitis) is the major insult limiting long-term survival [1,7,22,66–68], whereas the vascular disease is generally considered to be of secondary importance [1,7,22,66,67]. In the liver, both bile duct loss and OA (see below) together contribute to allograft failure [69–73]. Nevertheless, there are many similarities in the various manifestations of CR between solid organ allografts [63,74,75], which enable us to initially cover the topic from a common pathophysiological perspective. The organ-specific peculiarities that exist, often arise as a result of physioanatomic variations, some of which in turn, can influence the immunological mechanisms responsible for the injury.

Common features of chronic rejection

Introduction

In this section, we will review the common findings of CR in various solid organ allografts, which include: 1) graft vascular disease or obliterative arteriopathy(OA); 2) lymphohistiocytic interstitial inflammation; 3) patchy interstitial fibrosis, 4) destruction and atrophy of parenchymal cells; 5) destruction and atrophy of organ-associated lymphoid tissues and lymphatic vessels.

Graft vascular disease or obliterative arteriopathy (OA)

Distribution of the Lesions

The characterization of OA as a diffuse concentric intimal thickening involving the entire intra-organ arterial tree [63,64,74–76] is generally helpful in distinguishing it from atherosclerosis, which is endemic in the general population. However, in an individual lesion, it is not always possible to distinguish between the two disorders [60,77,78]. This is particularly true for organs such as the heart that are normally prone to atherosclerosis, but less of a problem in athero-resistant organs, like the liver. In an effort to avoid confusion, “obliterative arteriopathy” (OA) will be used throughout this review to refer to allograft arterial disease associated with rejection, whereas the term “atherosclerosis” (AS) will be used for the common endemic disease, realizing that the distinction is less than perfect.

The distribution of OA lesions in allografts is one of the key features used to distinguish it from AS. Whereas AS preferentially affects the large extracorporeal arteries [1,63,64,74], it has been repeatedly shown that OA more often involves both the extra-organ (e.g., epicardial, hepatic hilar, etc.) and first, second and third-order medium-sized intra-organ muscular arteries [1,63,64,74–76]. However, it has been our experience that OA lesions are not as diffuse as one might expect from a review of the pathology literature [1,63,64,74–76]. Even though smaller arteries are involved, the lesions in both large and smaller arteries are often patchy in distribution, and it has also been our impression that they often begin and evolve more quickly near branch points [63].

Small arteries (40–50 µm in external diameter) and arterioles commonly sampled in liver, heart and intestinal allograft biopsies can also be involved with OA; but it is much less common than large and medium-sized vessel disease [63,79] and the changes seen are less specific for the diagnosis of CR [63] Medial hypertrophy and/or mural hyalinization and eventually, frank destruction of the microvasculature, have been described in renal and liver allografts with CR [69,80,81]. Misorientation and compartmentalization of medial smooth muscle cells and adventitial fibrosis have been described in small subendocardial arteries from cardiac allografts [79], but these lesions do not necessarily correlate with the presence or absence of OA.

Arterial inflammation

Because OA is so strongly correlated with an alloimmune response, it is tempting to conclude that arterial inflammation causes the damage by recognizing normally expressed or upregulated class I and II donor MHC antigens on the endothelium [63,82–86]. However, arterial inflammation is not always observed [87,88] and fibrointimal hyperplasia can progress without it [88–92]. When inflammation is not seen, it is thought that antibody alone initiates the damage, or that the stereotypic vascular repair response can proceed without continued immunologic injury [89,92].

Even though OA can apparently occur without inflammation, most investigators would concede that cellular immunity importantly contributes to its development in many cases. The preferential localization of leukocytes in the adventitia and intima suggests that these are the most important antigenic targets or sites of damage [63,75,82–85,93]. The adventitia is rich in lymphatics and donor dendritic cells making it a site of both peripheral sensitization and a conduit for emigrating leukocytes in acute rejection [63,82,85,87,93,94]. When rejection is mild, the inflammation is usually limited to the adventitia [87]. When it is unusually severe, or when the recipient harbors anti-donor antibodies, mononuclear and/or neutrophilic endothelitis, respectively, are usually also present [74,77,82,95–101]. In many cases the intimal inflammation is quickly followed by fibrointimal hyperplasia, which marks the beginning of OA.

The inflammation often persists in OA lesions showing fibrointimal hyperplasia. At this stage, the lymphocytes present might be recognizing non-MHC antigen(s), like heat shock protein [102]. In any event, it is likely that toxic effector molecules and cytokines released by the inflammatory cells contribute to disruption of arterial wall homeostasis and arterial remodeling, since the two often co-exist [63,82–84,86]. However, the inflammatory cells often become separated from the endothelial layer by a concentric ring of intimal myofibroblasts (fig. 1). Thus, the inflammation is no longer directly associated with the endothelium. Instead, it forms a ring in the deep intima, which is separated from a cuff of adventitial macrophages by the media. The two inflammatory cell populations are connected with each other via a stream of individual inflammatory cells that permeate between vacuolated medial myocytes. This suggests that dynamic trafficking occurs between the adventitial and the deep intima, while the endothelium remains intact.

Fig. 1.

a) This section shows a hepatic artery branch in the hilum of a liver allograft. It is stained with a double immunolabeling procedure for smooth muscle actin(red) and macrophages(CD68: brown). Note the severe lumenal narrowing because of obliterative arteriopathy and destruction of the media along one half of its circumference(A = adventitia: M = media: I = intima). Note also the concentric ring of myofibroblasts(arrow) that separates the deep intimal macrophages from the endothelium, which is intact and uninflammed. Extension of the macrophages from the adventitia through to the deep intima suggests that some of the arterial inflammation may begin in the adventitia and may contribute to the development of OA without directly involving the intima. b) A higher magnification of the subendothelial ring of myofibroblasts shows them to be micotically active{arrow: double staining for smooth muscle actin (red) and proliferating cells. Ki-67 (brown)}. c) The ring of deep intimal inflammatory cells is also mitotically active{double staining for macrophages (CD68: red) and proliferating cells, Ki-67 (brown)}. d) Double labeling for smooth muscle cells(actin; brown) and male cells (Y chromosome in situ hybridization, yellow dot) shows that the intimal myofibroblasts in the female recipient of a male liver are of donor origin.

Immunophenotypic analyses have shown that the arterial inflammation consists primarily of an admixture of T-lymphocytes and macrophages [60,63,82,83,86,103]. In some studies, CD8+ T cells are the most common [83,103], a subset of which show perform positivity, identifying a cytolytic effector pathway [83,86] CD4/CD8 double negative and γδ T cells have been cultured in vitro from affected vessels [104]. The presence of occasional dendritic cells, signals ongoing antigenic presentation [82]. Macrophages however, often become the predominant inflammatory cell population, which may be related to the increased deposition of ground substance and lipid trapping, both of which stimulate phagocytosis(see below) [60,63,82,86,103,105–107]. The macrophages permeate the adventitia, media and intima, and proliferate within the artery [63]. Foam cell transformation is more common in early lesions and most often seen in liver allografts [60,63,72,74,75,105–108]. An adventitial-to-intimal macrophage size gradient is also often observed: those nearest to the lumen have abundant foamy cytoplasm, ones in the adventitia are smaller. Other inflammatory cells, such as dendritic cells [82], B cells [63,82], eosinophils [109] and plasma cells are much less commonly observed [63,82].

Mural histopathology

In practical terms, graft vascular disease can be equated with obliterative arteriopathy (OA), because vein involvement is both significantly less common and less severe [66,74,108]. The reason for preferential arterial targeting is uncertain, but hemodynamic factors, the prototypic arterial response to injury [110–112], endothelial antigenic differences and unique mechanical factors related to lymphatic disruption [87], are possible explanations.

It seems clear that OA evolves from a repair response to vessel wall injury, and all three layers of the artery wall are involved: intima, media and adventitia [63,74,111,113]. Documentation of the sequential changes have largely relied on a reconstruction of events from failed human allografts and autopsy specimens, and from detailed sequential analyses of experimental animal models. Common early findings include endothelial activation, which histologically manifests as hypertrophy or a “hobnail” appearance with eosinophilic transformation of the cytoplasm [63,85,87,114]. This impression is supported by the appearance of functional endothelial activation [114,115], including upregulation of class II MHC and adhesion molecules [63,87,116–120], and damage that can be objectively documented via a Fas-based apoptotic pathway [86]. In turn, endothelial damage results in loss of barrier function and the influx of clotting proteins(including fibrin), platelets, blood cells and lipids, all of which disrupt intimal homeostasis and contribute to the repair response and vascular narrowing(see below). Edema and an increased deposition of intercellular matrix and lipid [105], and a local turnover of cells [86,87] are seen in all three layers, but are especially common in the intima, which results in recruitment of medial myocytes and infiltration of foamy macrophages [74,77,82,95–100]. This is followed by intimal myofibroblast proliferation, that eventually replaces the entire intima. Breaks in the internal elastic lamina are variably present [63].

In the media, early changes include edema, and degeneration or frank necrosis of individual myocytes [87, 93]. Later, the media can become thinned [63, 81], presumably as a result of intimal migration of medial myocytes in response to injury. In fact, the media can be completely replaced by foam cells and/or fibroblasts so that the artery is no longer a flexible muscular conduit, but a rigid or loosely-wall tube formed by fibroblasts or foam cells, respectively.

Changes in the adventitia of arteries have received far less attention than they deserve. Early after transplantation, the adventitia is a primary site of sensitization, which results in inflammation and edema [87,94]. The adventitial inflammation often persists [63,82,87] and is followed by adventitial fibrosis [79,87] in the fully developed lesions. Ultimately, disruption of arterial wall homeostasis and lumenal narrowing predisposes to thrombosis and organ ischemia. Prototypic events are reconstructed in fig. 2.

Fig. 2.

Schematic of the sequential steps in the development of obliterative arteriopathy, which is the common feature of chronic rejection in all solid organ allografts. A normal artery is shown in frame 1, illustrating the important anatomic features including the endothelial lining (endo), which is flattened and inactive-appearing, the internal elastic lamina (iel), the media (med) and the dendritic cells(dc) located near draining lymphatics in the adventitia of arteries. Frame 2 shows early arterial changes during acute rejection and two distinct, but not mutually exclusive pathways of arterial injury. In severe acute rejection(a) the endothelium and intima can be directly injured by antibodies/complement and/or inflammatory cells, which leads to an intimal repair response and myofibroblast proliferation. However, arterial damage can also occur in less severe rejection episodes, illustrated by pathway (b). Inflammatory cells accumulate near adventitial donor dendritic cells and draining lymphatics. This causes edema of the media and intima with damage of individual medial myocytes, and indirect injury of the endothelium. Endothelial hypertrophy and activation occur with both insults and the arterial repair response is triggered. Frame 3 illustrates some of the latter charges in the development of OA. A concentric ring of myofibroblasts develops immediately subjacent to the endothelium, which returns to a more quiescent appearance. The media also becomes thinner, presumably as a result or migration of myocytes to the intima and/or compensatory arterial dilatation. The macrophages and lymphocytes now accumulate in the deep intima and focally stream through the disrupted media. This results in a connection between the intimal inflammatory cells and the cuff of macrophages and lymphocytes in the adventitia. The arrangement suggests that adventitial to intimal trafficking of immune cells is an important entry route. It occurs at a time when fibrosis is developing in the adventitia around affected anteries and in the interstitium.

Lipid and glycosaminoglycan deposition

OA is not commonly thought of as a lipid-dependent disorder, but there is evidence to suggest that hyperlipidemia can hasten its development [59,61,62,97] and measurement of the lipid content of arteries affected by OA show increased mean total cholesterol, esterified cholesterol, free cholesterol, and phospholipid content and concentration [121]. However, grummous atheroma, calcification and cholesterol clefts are less frequently observed in OA than in AS. McManus and colleagues [105,121] compared the patterns of lipid and glycosaminoglycan deposition in arteries affected by OA to that seen in AS in the general population. The lipoproteins seen in OA consist primarily of Apo(a) and apoE, which is in contrast to AS, where apoB is more prominent. Also, the apoE deposits can be seen in endothelia and in the extracellular space of the superficial intima with mild disease early after transplantation, and in the deeper intima with severe OA [107]. Co-localization of glycosaminoglycans shows that biglycan is particularly prominent in the intima of evolving OA lesions, but not in AS. Decorin is present mainly in adventitia of all vessels and in the intima of AS. Prominent versican accumulation is seen in the intima and media of arteries with OA, associated with smooth muscle and foam cells and the intimal biglycan and versican deposits positively correlate with the extent of luminal narrowing in OA [106]. Interestingly, the areas of Apo(a) and apoE deposition are also the sites of prominent proteoglycan sediments, especially versican. McManus, et al hypothesize that arterial injury, either mechanical and/or immunologic, leads to dramatic increases in intimal lipoprotein permeability and increased lipid retention within the arterial wall. This presumably occurs because the lipid interacts with chondroitin sulfate proteoglycans, including versican, which in turn, results in accelerated lipid endocytosis by macrophages and smooth muscle cells [107], and foam cell transformation.

Interstitial inflammation

Chronically rejecting liver, heart, intestinal and pancreatic allografts often show very mild and patchy inflammation, whereas in renal allografts, it can be more pronounced because of withdrawal of immunosuppression prior to allograft nephrectomy. The inflammation consists mostly of lymphocytes, macrophages and plasma cells. Formation of nodular lymphoid aggregates, some of which also contain germinal centers, suggests ongoing antigen presentation and continual stimulation [87,122]. The aggregates are often located in the adventitia of arteries, and near the serosa or capsule of organs [87,122,123], which are also the site of draining lymphatics. Inflammatory cells are also individually scattered throughout the interstitium, where they are seen in close association with parenchymal cells undergoing damage and atrophy.

Immunophenotypic studies have shown primarily CD4+ and CD8+ T cells, occasional eosinophils [109] and macrophages [87,122,124–127], the latter which can be quite prominent. Compared to acute rejection, there are usually more B cells [116,127,128], and formation of B cell follicles is not uncommon. Increased transcripts for granzyme B; the presence of Th1-like cytokines IL-2 and IFN-γ [126] and the prominence of macrophages, suggests that CR is mediated by an immune response tilted toward the TH1 pathway.

Damage and atrophy of parenchymal cells and interstitial fibrosis

Epithelial cells which rest on a conventional basement membrane, particularly those that line conduits used for exchange of substances with the environment (e.g., bronchioles, bile ducts, pancreatic ducts, renal tubular epithelium), are particularly prone to damage during CR. There are at least several possible explanations for this observation, which are covered in greater detail in a recent review of the immunopathology of liver allograft rejection [129]. Basement membranes contain matrix components important to the migration, positioning and co-stimulation of T-cells [129]. In addition, epithelial cell “immunogenicity” is enhanced by unique antigenic profiles [129] and their ability to upregulate immunologically active adhesion and co-stimulatory molecules and to produce cytokines and growth factors [129,130]. Since the epithelial-lined conduits also transport antigens to and from the environment, they are often intermixed with antigen presenting cells, and surrounded by a rich lymphatic plexus (see below), which drains to the regional lymph nodes. This is normally a site of local immune activation to environmental and microbial antigens, which in an allograft, could trigger rejection. Lastly, the bile ducts and bronchioles are almost exclusively supplied by an arterial circulation, obliteration of which can cause ischemic injury. Thus, both direct immunologic injury and ischemia likely contribute to the damage and loss of small bile ducts and bronchioles during CR [69.131].

Patchy interstitial fibrosis is another hallmark feature of CR, but alone, it cannot be used as a reliable predictor of CR in peripheral biopsy samples [132]. Small scars are frequently seen as a result of previous biopsies or other insults. Nevertheless, fibrosis is a constant feature of CR, and it often manifests as expansion and collagenization of the connective tissue normally surrounding the arteries affected by OA. This is combined with more localized interstitial widening [63], especially in areas of ongoing immunologic damage, which are often associated with the deposition of tenascin and other matrix components [133,134], endothelin [135] and activation of interstitial myofibroblasts [134]. Larger scars representing healed infarcts and fibrosis in watershed regions supplied by the terminal circulation, suggest that ischemic injury also contributes to fibrogenesis [63].

Disruption of lymphatics and organ associated lymphoid tissues

Complete revascularization of an allograft results in the circulation of recipient immune cells through donor organ-associated lymphoid tissues(GALT, BALT, portal lymphoid tissue) [136–140] and regional donor lymph nodes [141]. Early after transplantation, this results in a “in vivo mixed lymphocyte response”, which typically manifests as acute rejection [136–139,141]. Transplantation also disrupts the efferent lymphatics, which results in organ edema and contributes to the re-implantation response. The lymphatic channels reconnect within two to three weeks [142, 143] in the absence of additional insults. Acute rejection however, results in the increased production of lymph fluid and at the same time, disrupts the lymphatic microvasculature, both of which contribute to the reappearance of graft edema and swelling during rejection reactions [142,144–147]. In CR, the organ-associated lymphoid tissue, such as the BALT [147,148] or GALT [139], is destroyed and replaced by fibrosis. Patchy interstitial fibrosis also focally disrupts intra-organ lymphatics [87, 149]. Both of these changes undoubtedly contribute to an inability of chronically rejecting allografts to adequately process infectious agents and antigens that are normally cleared via these pathways [8,37,40,148,150–152]. In addition, disruption of lymphatic drainage alone has been shown to produce arterial injury, which resembles changes seen with developing OA [153]. Thus, we [87] and others before us [149], have suggested that this potential mechanism of arterial injury might importantly contribute to the development of OA.

The direct injury model of pathogenesis

Cause(s) of the arterial injury

Animal models are an excellent way of controlling experimental conditions to isolate the various arterial insults. Many models that study immunological injury ameliorate acute rejection, so the allograft survives long enough to develop CR and OA. One approach exploits a weak immunological barrier, where only minor histocompatibility barriers are crossed(e.g. LEW to F-344 rats) [95–99,154,155]. Another approach uses transient treatment with various forms of immunosuppression with the most popular being a short course of anti-CD4 monoclonal antibodies [85,115]. The percentage of animals surviving long term and developing OA can be adjusted by titrating or adding immunosuppression, depending on the model.

In these models, there is relentless direct immunological injury to the allogeneic arterial endothelium from prolonged, non-lethal acute rejection, that appears to be mediated primarily by T lymphocytes, macrophages and anti-donor antibodies [85,95,98,99,114,115,154,156]. This quickly evolves into CR and OA, much like the clinical cases that directly evolve from an inadequately controlled acute rejection. Indeed, on the basis of these models and clinical studies [63, 75, 77, 157], many investigators would support the following pathogenesis of OA: direct lymphohistiocytic inflammation of allogeneic endothelium or alloantibody binding → endothelial cell damage → release of cytokines and growth factors by endothelial cells, myofibroblasts and inflammatory cells → disruption of intimal homeostasis → intimal myofibroblast proliferation → lumenal narrowing. For the purposes of discussion, this will be called the “DIRECT INJURY HYPOTHESIS”.

Substitution of immune deficient and/or knockout mice as recipients into these schemes, has further clarified the nature of the immunological injury by isolating the contribution of humoral and cellular components [88,91]. The development of OA in combined immune deficient mice treated with repeated injections of exogenous anti-donor antibodies [88] nicely illustrates the importance of antibodies. Other models also clearly show a role for antibody-mediated damage [158,159]. However, OA also develops in B cell knockout mice that presumably are incapable of an antibody response [88,91], showing an important contribution by cellular immunity. Other insults that could contribute to the arterial injury or repair response include preservation/peri-operative ischemic injury [160,161]; cytomegalovirus infection [34,64]; altered mechanical or hemodynamic forces because of either decreased renal nephron mass [6], or immunologically-mediated lymphatic disruption [87]; and other well-known atherogenic insults such as hyperlipidemia, hypertension, diabetes and smoking.

Mechanisms of myofibroblast proliferation

Experimental animal models have also nicely shown that the initial arterial insult may be immunologic(antigen dependent), but the stereotypic repair response that follows does not appear to require the persistent influence of alloimmunity at some stage in its development(antigen independent) [89, 90, 92]. Thus, regardless of the source or mechanism of injury, the response of the vessel wall is the final common pathway for the development of disease and ultimately, organ dysfunction. This has led to a natural pre-occupation with understanding the molecular mechanisms of myofibroblast proliferation [162]. To study this response, a number of investigators use the isolated aortic allograft model, introduced by Häyry [19]. Although it lacks the organ parenchyma [92, 94, 163–166], this model offers a quick way of screening agents that might be of potential therapeutic benefit(Häyry, personal communication).

Research to date has implicated multiple and redundant signaling molecules that initiate and/or promote smooth muscle cell proliferation and fibrogenesis [19,97,162,167–171], similar to the situation observed with AS in the general population [172–175]. This includes growth factors, cytokines and chemokines [17,19,97,118,156,162,167–171,176–180]. For example, it has been shown that platelet-derived growth factor A and B [167,168,181], fibroblast growth factor [167,170,182], insulin-like growth factor-1 [162] and interleukin-1[162] can be released by myofibroblasts, endothelial and inflammatory cells into the arterial intima and all of them can stimulate smooth muscle cells to migrate to the site of injury and proliferate. Macrophages appear to be an important source of molecules causing damage, and initiating the repair response and fibrogenesis [17,97,118,126,154–156,176,177,183]. This helps explain the inhibition of OA by an essential fatty acid deficient diet, which interferes with leukocyte chemotaxis [97]. Transforming growth factor-beta 1 might importantly contribute to fibrosis [134,167,169,171,179]. Interleukin-6 might has been linked to the glomerular proliferative lesions in chronic renal allograft rejection [118]. The reader is referred to several excellent reviews for a more detailed accounting of this complex area [19,21,161,162,167,172–174,183,184].

Manifestations of chronic rejection in specific organs

Liver allografts

Chronic liver allograft rejection usually is not seen before 2 months [185] after transplantation and most frequently develops after: 1) an unresolved episode of acute rejection; 2) multiple episodes of acute rejection; or 3) indolently over a period of months to years, with few or no clinically apparent acute cellular rejection episodes [27]. Risk factors have already been discussed. Often the only reliable early indicator of CR is persistent and preferential elevation of γ-glutamyl transpeptidase and alkaline phosphatase [27, 186], which is related to bile duct damage and loss [26,27,69,71,73,187–191]. If clinical symptoms are present, they usually resemble those of acute rejection, until allograft dysfunction becomes severe enough to cause jaundice [26,27,69,71,73,187–191]. Biliary sludging or appearance of biliary strictures, hepatic infarcts, and finally loss of hepatic synthetic function, which can manifest as coagulopathy, malnutrition, and hepatosplenomegaly [27]are late findings presaging allograft failure.

Chronic liver allograft rejection is histopathologically defined by two main features: severe damage and loss of small (< 60 µm) bile ducts and, as in all other solid organs, OA [26,27,69,71,73,187–191]. Cases with either isolated bile duct loss or foam cell arteriopathy alone may occur, but usually both features are found together [72, 73]. Unfortunately, arteries with pathognomonic changes are rarely present in needle biopsy specimens, and therefore what has been observed about OA in the liver has come from examination of failed allografts removed at the time of re-transplantation. Therefore, in biopsy specimens, significant weight is usually placed on damage and loss of small bile ducts, but ductopenia can be non-specific finding that occasionally occurs as a result of non-rejection-related complications [192].

As with CR in other organs, the inflammation is often inconspicuous, consisting of lymphocytes, plasma cells and macrophages [116, 126, 193]. Despite a paucity of portal inflammation, intra-epithelial lymphocytes are located adjacent to degenerating biliary epithelial cells [26,27,69,71,73,187–191], which show uneven spacing of individual epithelial cells, eosinophilic transformation of the cytoplasm and ducts only partially lined by bile duct cells (fig. 3) [194]. The portal infiltrate is T cell predominant in CR and contains both CD4 and CD8 subsets [116,126,193], with the latter predominating in and around the biliary epithelium [193]. Increased γ-interferon, IL-2, granzyme B transcripts [126] have also been detected, the last of which strongly suggests that a cytolytic effector pathway of injury is involved. The damaged ducts can also show "inappropriate" expression of the major ABO blood group antigens [195], the significance of which is uncertain. The diagnosis of CR at these “earlier” stages [192, 194, 196–198] can be difficult. Review of previous biopsies and clinical correlation typically show a patient with documented acute rejection, that has progressed to chronic injury and prolonged liver dysfunction, which has not responded to appropriate anti-rejection therapy [27]. The diagnosis of CR can also be supported by selective hepatic angiography that shows pruning of the intrahepatic arteries with poor peripheral filling and segmental narrowing [26,191,199,200].

Fig. 3.

a) This section some of the most characteristic changes associated with the earlier phases of chronic liver allograft rejection. Note the “atrophic” appearance of the biliary epithelium(arrow). There is uneven spacing of the epithelial cells, eosinophilic transformation of the cytoplasm and local pyknosis. b) Interstingly, even though these duccal cells are “atrophic” appearing, they can show an increased proliferation rate, as determined by Ki-67 staining(arrow): c) Eventually, the small bile ducts are completely destroyed in chronic liver allograft rejection. Cytokeratin staining can be used to help identify the ducts, but it is usually not needed [288]. Whether these few periportal cytokeratin-19+ cells in this chronically rejected liver(arrows), represent progenitor cells capable of regenerating the ducts, is worthy of further study(see text).

Interestingly, despite the destruction of small ducts, a ductular reaction, or compensatory ductal proliferation in the marginal zone is not generally seen, unless the patient recovers. In such cases a “ductular reaction” has been described before clinical improvement [71,201,202]. Whether the recovery of ducts is related to proliferation of residual ductal epithelium, maturation of liver progenitor cells, or metaplasia of periportal hepatocytes, is a intriguing question about liver growth and development [129].

The portal tract connective tissue assumes a “scarred” appearance by routine light microscopy, which is related to a destruction of the peribiliary plexus [80]. An increase of extracellular matrix proteins fibronectin, tenascin, undulin, and collagen VI [134] have been described in the perivenular regions and this is associated with an accumulation of alpha-actin-positive cells and TGF-beta1-expressing macrophages in the areas [134].

Since the presence of foam cell arteriopathy can rarely be confirmed by a core needle biopsy, other "surrogate" features are often used to suggest the presence of OA. Features that suggest but do not prove the presence of foam cell arteriopathy include loss of arterioles and small (<20 µm) portal tract arteries, centrilobular hepatocellular swelling, perivenular sclerosis, and centrilobular hepatocyte dropout [26,27,69,71,73,187–191].

Lobular changes commonly seen in CR include centrilobular hepatocanalicular cholestasis, intra-sinusoidal foam cell clusters, mild spotty acidophilic necrosis of hepatocytes, centrilobular hepatocyte atrophy and/or ballooning and perivenular sclerosis [26,27,69,71,73,187–191]. The centrilobular degeneration and perivenular sclerosis may be related to either ischemia or damage from repeated bouts of “central venulitis” during acute rejection.

As with other organs, the diagnosis of CR is easier to establish in an explanted failed allograft. The presence of significant foam cell OA, should be seen in at least some of the muscular arteries in the hilum [26,27,69,71,73, 187–191]. Major hilar bile ducts may show sloughing of the epithelium, focal necrosis, intraluminal papillary hyperplasia of the epithelium, mural fibrosis and acute and chronic inflammation [203].

There are two fascinating aspects of chronic liver allograft rejection: its potential reversibility [71,202], which has already been discussed; and an appreciably lower incidence than other allografts (table 1, [51]). Experimental animal studies have also shown that a liver allograft can also protect other organs from the same donor from CR [87].

Resistance of liver allografts to CR can best be explained by the immunological theories of so called “hepatic tolerogenecity”. They can be broadly separated into two general categories, based on whether emphasis is placed on the parenchymal or non-parenchymal portion of the liver, which in turn, results in: 1) deletion or 2) functional anergy in the responding T cell repertoire. Release of soluble donor MHC class I antigen from the allograft, supporting the importance of the parenchyma, does not seem to be a tenable hypothesis [204]. Murine liver allografts are routinely accepted between strains of mice that show no difference between the class I loci, but are mismatched for class II [205]. In addition, fully allogeneic liver allografts from class I or II MHC deficient mice, which do not shed soluble MHC antigens [205,206], are also accepted. Moreover, other organs also secrete soluble MHC antigens [207] but they are routinely rejected, and infusion of exogenous soluble MHC usually leads to only slight graft prolongation.

One potential explanation for the importance of the parenchyma relies on the concept that allogeneic hepatocytes provide only one of two signals needed for allogeneic lymphocyte activation [208]. Lack of important co-stimulatory molecules is thought to result in the induction of anergy in the responding lymphocyte populations [208]. However, when transgenic mice that aberrantly express the allogeneic MHC class I molecule H-2Kb (Kb) in the liver, using a metallothionein promoter [209], are crossed with a strain that develops CD8+ T cells specific for the Kb molecule, lymphocyte-mediated “autoimmune” liver damage can be induced under certain conditions. Interestingly, most of the CD8+ cells responsive to Kb were eliminated by the intense intrahepatic activation, but some of the liver continued to show chronic low grade inflammation [209]. However, great care had to be taken to assure that alloantigen expression was limited to the liver [209]. Interpretation of these complicated experiments is difficult, but it appears that the liver may be able to effect an activation-induced partial clonal deletion of allogeneic lymphocytes [209], as originally proposed by Kamada [210]. This may represent a special circumstance, because alloantigen expression on other non-hematolymphoid cells such as pancreatic islets, does not lead to anergy [211], but to immunological “ignorance” [212], which is an important difference. Induction of true anergy may in fact, require presentation of antigen by professional antigen presenting cells.

It has been our contention that so called “hepatic tolerogenecity” is in large part, a result of hematolymphoid microchimerism [48,213,214] sustained by donor hematopoietic stem contained within the liver [215–217]. The immunologic mechanisms involved in graft acceptance is initially, a non-deletional, active process that appears to require the presence of donor APC [87,218]. In microchimeric recipients, the donor cells cause low grade proliferation and activation of the recipient immune system [51,87,218–220], including activation of B cells [219,221] and alterations in IL-4 production [222]. This in turn, leads to hyporesponsiveness to donor antigens and graft acceptance.

A recent series of investigations by Bishop et al [223–226] add credence to the viewpoint that hematolymphoid cell-induced activation of the recipient is important in liver graft tolerance. Comparing spontaneously accepting and rejecting rat strain combinations, they have shown that acceptance of the original liver allograft and subsequent grafts from the same donor, resides with a population of radio-sensitive hematolymphoid cells within the liver [226]. Destruction of the hematolymphoid cells by irradiation or blockage of the response by a short course of methylprednisolone reduces survival of subsequent donor strain skin grafts [226].

Heart

Obliterative arteriopathy is the major manifestation of CR in cardiac allografts. Unfortunately, cardiac allografts are also susceptible to AS that is endemic in the general population. This results in the frequent transmission of coronary artery disease from the donor to the recipient [56–58]. Thus, AS and OA both contribute to the arterial disease seen in allografts before and after cardiac transplantation [58,227–230].

Cardiac dysfunction associated with CR is largely because of ischemic damage to the heart. However, the classical symptoms of myocardial ischemia like angina pectoris, are thought not to be reliable indicators in the heart transplant recipient, since the allograft is denervated [3]. Nevertheless, in a series of 25 myocardial infarcts in heart transplant recipients [231], symptoms included chest pain (n=2), arm pain (n=3), weakness (n=16), dyspnea (n=11) and palpitations (n=8). Three episodes were clinically silent, detected only as new electrocardiographic changes during routine follow-up and two patients had old Q waves. Of the 8 patients hospitalized with symptoms, only 7 had typical Q-wave changes on electrocardiography; 5 had nonspecific ST-segment changes and 2 had no documented changes. Serial creatine phosphokinase levels were obtained in 13 patients, and values were elevated in 8. Other manifestations of CR and myocardial ischemia include cardiac enlargement with ventricular dilatation; loss of papillary muscle function and mitral regurgitation; ventricular wall dysfunction [3], constrictive pericarditis [232] and the development of congestive heart failure and/or cardiogenic shock, which can be misdiagnosed as infection [231].

One might expect that pre-existing coronary artery AS would accelerate the development of OA. Indeed, the use of older donors has been associated with a higher incidence of vascular abnormalities early after transplantation [57,229,233]. However, pre-existing donor artery disease does not necessarily impart an increased risk for the development of OA [57,234], which developed at different sites in the vessels. This suggests that at least some of the pathogenic mechanisms responsible for OA are different from those involved in AS. Clearly, more detailed investigation into the association between OA and AS is needed.

Some of the cardiac pathology can be detected by echocardiography or serial radionucleotide angiography before classic symptoms and signs of heart failure arise [3]. Although these methods of observation are less sensitive than histopathological examination of the entire allograft [235], they are more reliable than evaluation of endomyocardial biopsies, which can rarely detect OA (see below).

Gao et al. [236] originally classified the angiographic abnormalities into three categories: type A, discrete or tubular stenoses; type B, diffuse concentric narrowing; and type C, narrowed irregular vessels with occluded branches. Use of this classification or something similar, has proved to be a useful approach when applied to allograft coronary artery pathology. Type A lesions of the primary epicardial coronary vessels. with less frequent secondary branch involvement are most commonly encountered in non-allograft hearts. Not unexpectedly, the same type of lesions are detected in the allograft population shortly after transplantation, consistent with transmission of donor disease. However, type A lesions involving secondary and tertiary branches develop more frequently in allografts than in native hearts. In addition, type B and C lesions (diffuse narrowing, tapering and obliteration), are much more common in allograft coronary arteries than in native hearts and involve primary, secondary and tertiary branches [236]. Total vessel occlusion is found predominantly in the proximal or middle vessel segments of atherosclerotic disease, whereas distal involvement can be seen in up to one half of patients with OA. Failure to develop collateral vessels is also more common in allograft recipients with OA. Based on these observations, Gao et al. [236] concluded that coronary artery disease in transplant patients represents a mixture of typical AS and unique transplant-related progressive distal obliterative disease that occurs without collateral vessel development [236].

Several investigators have drawn the same conclusion using the more recently developed intravascular ultrasound [58,227–230]. This technique offers some advantages over angiography in terms of sensitivity [233,237,238] and the ability to more precisely characterize changes in coronary vessel shape and wall thickness [239], which enables one to follow the changes over time. In general, lesions frequently detected early after transplantation reflect donor AS in the general population: segmental and eccentric intimal thickening and lumenal narrowing, primarily involving the epicardial coronary arteries and their primary branches. Over time, more distal, diffuse, non-calcific and concentric intimal hyperplasia develops [58,228,230,238,240], which likely represents OA.

Serial intracoronary acetylcholine infusions and quantitative angiography have concluded that endothelial cell dysfunction is frequently observed in allograft coronary arteries and it often occurs early after transplantation [241,242]. A close association between endothelial dysfunction and intima abnormalities can be documented in some [241] but not all studies, presumably because of the episodic nature of the immune injury [227]. Other abnormalities include an impaired responsiveness of epicardial arteries to increased flow [242].

Except for rare cases [243], endomyocardial biopsies do not show changes that can be reliably used to make the diagnosis of CR [64,65,244–248]. Intimal thickening does not usually affect the small penetrating intra-cardiac arterioles present in endomyocardial biopsies [79], and collagen density in biopsy specimens does not appear to predictably increase with time in patients who develop CR [132]. There is however, some evidence that nodular sub-endocardial lymphocytic aggregates or Quilty lesions [244] may be associated with OA.

Quilty lesions are present in 4–25% of human endomyocardial allograft biopsies and are detected on at least one occasion in 58–78% of patients [246,247,249, 250]. Although they are not the equivalent of therapy-requiring acute rejection, some clinical studies show an association with acute rejection in the underlying myocardium [247, 251–254] or with OA [248]. Patients in the Stanford series with Quilty lesions who underwent transplantation in the early 1980’s had the highest incidence of graft vasculopathy [123], but more recently the incidence of OA has significantly decreased in these patients [123]. This perplexing observation might be explained by the fact that these patients had the most potential for improvement, particularly with the addition of new immunosuppressive regimens. For example, Tacrolimus has significantly lessened the incidence of Quilty lesions in Our heart transplant recipients [255] and significantly increased the half-life of kidney allografts [256], presumably by lessening CR.

Even though a causal association of Quilty lesions with OA is unlikely, these observations suggest that Quailty lesions might not be innocuous histopathological curiosities. Their presence in animal models with persistent alloactivation and developing OA [87] suggest that they serve as a signal of indolent ongoing immunological damage, which has the potential to contribute to the development of OA. Kemnitz et al. [253] offered an explanation for the possible association between Quilty lesions and OA by speculating that they may represent form of vascular rejection, since the heart develops embryologically from primitive vessels.

Consistent with an immune etiology [1,59,65,83,235, 245,257], it has been our experience in humans, that pure OA frequently involves both smaller intramyocardial arteries and epicardial coronaries and infrequently shows eccentrically situated and grummous atheromas and calcifications (fig. 4). In addition, the fibrosis in CR often extends along the connective tissue septal containing the affected arteries. This is in contrast to AS, that tends to show proximal epicardial coronary involvement by eccentric lesions with grummous atheromata and sparing of intramyocardial branches. OA can also clearly be shown to develop directly from Inflammatory and/or necrotizing arteritis is some cases [77,83,235], much like experimental animal models [85,87,95]. However, as already pointed out, even though OA involves both proximal and more distal portion of the arteries, OA is often not as diffuse or concentric as one might expect from a review of the literature [1,59,65,83,235,245,257]. In our experience, OA is often patchy in distribution (preferring branch points as epicardial vessels penetrate the myocardium), shows a significant adventitial components, can develop eccentrically, and frequently does not extend to the smallest arteries and arterioles present in endomyocardial biopsies [63, 79]. This is consistent with experimental animal models where the OA lesions are definitely a result of immune injury, but notoriously heterogeneous in distribution and severity [85,87,95]. In fact, the distribution is so heterogeneous that elaborate scoring systems are required to quantify the findings [85,87,95].

Fig. 4.

a) Atherosclerotic lesions involving the epicardial coronary arteries are frequently eccentric and show cholesterol clefts, as is seen in this lesion from a cardiac allograft. b) However, OA lesion can also have an eccentric distribution as shown here. Note the cuff of adventitial and deep intimal macrophages (arrows, CD68), similar to that seen in Figure 1 from a liver allograft. c) In contrast to acherosclerosis, OA often involves the distal branches of the epicardial coronary arteries and d) even those smaller intra-myocardial branches (arrows). There is also destruction of the lymphatic microvasculature of the heart in CR (e and f) Compare the lymphatic channels (black, histochemical stain for 5 nucleortidase) traversing the fibrous septae with arteries in a normal heart(e), with the lack of staining seen in cardiac allograft with CR(f)

However, in an individual lesion, the relative contribution of OA or AS may be difficult to determine. This is not surprising since atheromas are frequently inflamed with donor hematolymphoid cells and therefore would be more immunogenic than other constituents of the allograft. Conversely, McManus et al. [105,107,121] have shown the tendency for OA lesions to imbibe lipids. Thus, some histopathological studies of failed allografts and autopsy specimens have encountered some of the same difficulties as the imaging studies: Atherosclerosis and OA both contribute to the coronary artery disease seen in allografts. This might explain some of the more recent apparent inconsistencies in the distribution of OA noted by investigators using intravascular ultrasound [258].

Small intestine

Transplantation of the immunologically difficult small bowel, has only recently been clinically achieved to any significant extent [259–261]. This was largely because of ineffective control of acute rejection, which was made routinely possible by the use of FK506 (Tacrolimus) immunosuppression [262]. Thus, until recently, when the entire spectrum of human pathology was delineated [25], most of the CR changes small bowel allografts have relied on experimental animal models [139,140,263–265].

In experimental animals, CR of the small intestines often evolves from suboptimally controlled acute rejection [140]. It is characterized by OA, hyperplasia of the muscularis propria [266], apoptotic death of epithelial cells in the crypts of Lieberkuhn, villous blunting, focal fibrosis of the lamina propria, and destruction of donor mesenteric lymph nodes and the mucosal-associated lymphoid tissue(Peyer’s Patches) [140]. The last complication results in focal mucosal ulcers [139,263–267]. Although experience in humans is limited, detailed evaluation of the few chronically rejected human allografts [25] have shown the same set of histopathological changes as seen in the experimental animals (fig. 5).

Fig. 5.

a) CR of human small bowel allograft shows focal, non-healing ulcers(arrow), as shown here, and thickening and fibrosis of the submucosa(sm). b) A higher magnification of the submucosa reveals abundant foam cell deposition, similar to that seen in the arterial intima of blood vessels with OA. c) Like other allografts, the arteries affected by OA are not commonly sampled in biopsies. This section shows a mesenteric artery with OA near a mesenteric lymph node with lymphold depletion and fibrosis(arrow). d) A higher magnification shows mild lymphocytic intimal inflammation, intimal foam cell deposition and endothelial cell hypertrophy. Note also the marked medial vacuolization, which is due to foam cell infiltration between myocytes and intercellular edema.

As with other allografts, the biopsy diagnosis of CR in humans is difficult. The vessels affected by OA, primarily the distal branches of the mesenteric arteries and the larger arteries of the serosa and muscularis propria, are not routinely sampled. Therefore, one is forced to rely on “surrogate” findings. These include focal, non-healing ulcers; a distinctive combination of regeneration and evident crypt epithelial apoprosis; and a disordered mucosal and villous architecture with fibrosis [25]. Typical clinical findings that support the diagnosis of CR include, persistent diarrhea with non-healing ulcers, repeated bouts of suboptimally controlled acute rejection; and either non-compliance with immunosuppressive medications or deliberate withdrawal of immunosuppression because of life-threatening infections [25]. Endoscopic and radiographic studies used to support the diagnosis of CR include focal loss of mucosal folds, focal ulcers and mural thickening, and pruning of the mesenteric arterial arcades, respectively.

Incompleteness of the direct injury model

In addition to immunological arterial injury, a number of clinical and experimental observations suggest that other factors contribute to the development of OA [6,16,51,63,70,82](table 2). The Direct Injury Model also does not address the immunological mechanisms responsible for persistent allograft injury. This is probably related to an incomplete understanding of the mechanisms of graft acceptance and a natural pre-occupation with the molecular mechanisms of the control of smooth muscle cell proliferation and fibrogenesis. In our opinion, one mechanical factor that should receive more attention is the destruction of organ-associated lymphoid tissues and lymphatic channels, which alone can cause arterial pathology. Similarly, destruction of the vasovasorum, could indirectly injure the wall of larger vessels and trigger a repair response [1]. Regardless, any hypothesis of OA pathogenesis will have to accommodate the inadequately explained findings listed in Table 2. In addition, it may be helpful to view CR as incomplete tolerance, by comparing recipients with CR to genetically-identical controls who are tolerant to the graft, instead of to isografts controls, as is commonly done in many experimental systems.

Table 2.

Observations not adequately explained by the Direct Injury Hypothesis (see text)

Observation	Possible explanation(s)
Markers of severe acute rejection (alloantibodies or inflammatory arteritis), not always detected before development of OA.	-Severe acute rejection and inflammatory arteritis not required
Coronary arteries of composite heart-lung allograft recipients are less frequently and less severely affected than coronary arteries in isolated heart allografts (table 1).	-Mechanical factors involved -Antigen quality or quantity is of importance
Pulmonary arteries are less severely affected than coronary arteries in heart-lung transplant recipients [64]	-Mechanical factors involved -Antigenic or microenvironmental differences important
Inadequate explanation for progression of OA in absence of immunologic insult [17,89].	-Initial injury triggers self-perpetuating repair response, perhaps driven by normal physiology
OA is not a feature of graft versus host disease	-Mechanical factors involved in organ removal and implantation important

CR viewed as a “Incomplete Tolerance”

Implantation of any solid organ allograft results in a characteristic cycle of heightened immune activation, brought about the initial engagement of the donor and recipient lymphoid systems [48,214,218,268], followed by evolution toward a more stable relationship when immunosuppression can be considerably lowered or even withdrawn [269, 270]. In the past, this prototypic series of events was attributed to the presence of passenger leukocytes or donor hematolymphoid cells [271–273]. It was thought that they were essentially deleterious to allograft survival, and served only to directly sensitize the recipient [271–273]. When the donor cells were destroyed or died our, the donor/recipient immunological interface became less contentious. However, in some clinical [48,274–277] and experimental systems [87,278,279], the exact opposite was actually found: donor hematolymphoid cells persisted in successful long term allograft recipients, but were destroyed in recipients with CR. This phenomenon called “microchimerism” uncovered a paradox [218], resolution of which required a paradigm shift in transplantation immunobiology [214,218]. The new, Two-Way paradigm line of reasoning hypothesized that donor hematolymphoid cells are necessary, but alone, not sufficient for tolerance induction and thus, freedom from CR [48,87,213,214,218,278,279].

While attempting to study microchimerism, we inadvertently created a clinically relevant experimental animal model of CR [87,280]. In this model, genetically identical recipients were made CR-resistant or CR-prone by pretreatment with a donor liver allograft or bone marrow infusion, respectively [87], and then challenged with a cardiac allograft without immunosuppression. Both groups underwent a transient acute rejection within days after placement of the cardiac grates, but were operationally tolerant. The cardiac grafts were accepted and remain functional for more than 100 days. However, those pretreated with a bone marrow infusion were incompletely tolerant, because eventually they developed CR and OA, whereas those pretreated with a liver were resistant to CR.

A more in-depth analysis showed that the acute rejection crisis in the two groups could be separated on the basis of macrophage migration, cytokine production and alterations in MHC antigen expression. Acute rejection in the CR-prone animals was characterized by an influx and accumulation of recipient macrophages; persistent and strong upregulation of class II MHC antigens on the vascular endothelia; loss of intra- and extra-graft and donor hematolymphoid cells [87] and a cytokine mRNA response enriched in γ-interferon, IL-12 and macrophage activating chemokines(Murase, manuscript in preparation). In contrast, the strong upregulation of γ-interferon and IL-12 was not seen during acute rejection in the CR-resistant animals, even though there was evidence of graft inflammation by histology and immune activation by cytokine/chemokine mRNA analysis [87]. In addition, endothelial class II MHC expression in the CR-resistant recipients was weak and transient and donor hematolymphoid cells persisted in the graft and recipient lymphoid tissues.

Like other CR models, this one shows a clear link between CR and macrophage infiltration, which in turn, is likely related to the production of γ-interferon and other macrophage activating cytokines [97,177,180,267,281]. There is also evidence of persistent endothelial and smooth muscle cell activation, such as expression of MHC class II [87] antigens and adhesion molecules [17,85,97,114,115,118,155,156,163,176,241,282]. Unlike other models however, it shows that arterial remodeling starts to occurs during acute rejection before any intimal inflammation is seen. In addition, this model enables the investigator to compare and contrast completely tolerant recipients to those who are incompletely tolerant and develop CR. Major differences between these two groups are the persistence of donor hematolymphoid cells in the former, and heightened recipient macrophage influx in the latter [87]. Thus, our model more closely resembles the clinical cases that indolently develop CR.

The slower evolution of CR in some cases may be related to the mode of alloantigen presentation. As originally predicted by Batchelor [283], acute rejection is likely dominated by direct presentation of alloantigen by donor APCs, whereas CR may be primarily mediated via indirect alloantigen presentation by recipient APCs, consistent with the prominence of recipient APCs within the graft. Conversely, a bold and controversial hypothesis is that freedom from CR requires the persistence of donor hematolymphoid cells [48,87]. The idea is that to achieve tolerance, the donor immune system is required, since one of its normal physiological functions is to establish immunological boundaries, or the “self” [48,218]. Hopefully, the freedom from CR that is seen with augmentation of chimerism using harsh Conditioning regimens in experimental animals [284,285], will be realized in unconditioned human recipients as a lower incidence of CR [38,286,287]. Whether this line of reasoning is correct or not, only time will tell.