Many systems have been proposed for grading rejection histologically [1-13], but as yet none has achieved universal acceptance. Most grading schemes apply specifically to acute cellular rejection, which is the commonest form of rejection seen histologically. Some schemes incorporate features such as ductopenia , which might be regarded as more typical of chronic rejection and this may give rise to diagnostic problems. Such problems probably reflect our lack of understanding of the relationship which exists between these two patterns of immune-mediated graft damage.
In order for a histological grading system to be useful, it should fulfill the following 4 criteria: (1) Scientific accuracy - i.e. based on histological changes known to be important in causing graft damage; (2) Simplicity - both for the Pathologist assessing the biopsy and for the Clinician receiving the report; (3) Reproducibility and (4) Clinical utility in treatment and prognosis. As far as the final criterion is concerned, there are two main questions which need to be addressed. Firstly, does histological severity of graft rejection correlate with clinical and biochemical signs of graft dysfunction at the time of biopsy? In many centres (including Birmingham) additional immunosuppression is only given when there are both clinical and histological signs of rejection. Clearly it is desirable for these to be closely correlated. Secondly, can one identify features at an early stage (e.g. in a 7 day biopsy) which might predict subsequent poor outcome? In most cases this would mean progression to graft failure from rejection. Identification of such cases at an early stage would be also be useful as they may be candidates for switching to more potent immunosuppressive agents before irreversible damage has occurred.
In this paper I shall briefly review the main histological features of acute liver allograft rejection, describe how these have been incorporated into some European grading schemes and then discuss how the "European Approach" compares with grading schemes currently in use in the United States. There have been some limited attempts to grade chronic rejection histologically [4, 13, 14,] but these will not be considered further here.
(1) Portal Inflammation There is typically a mixed infiltrate which includes lymphocytes, other mononuclear cells, neutrophils and eosinophils. A prominent component of large "activated" blast cells is sometimes present. The presence of a mixed inflammatory infiltrate may be used not only in diagnosing acute rejection, but also in grading it histologically. In some cases where portal inflammation is intense there may be spillover into periportal regions. However, this feature is not usually marked.
(2) Bile Duct Damage Bile ducts are surrounded and infiltrated by a mixed population of inflammatory cells which usually includes a prominent component of neutrophil polymorphs . In some cases, there are luminal collections of pus cells, mimicking changes seen in ascending cholangitis. Biliary epithelium may show degenerative changes including cytoplasmic vacuolation, individual cell dropout  and focal disruption of the basement membrane. In some cases bile duct epithelium shows nuclear pleomorphism and hyperchromatism, producing a "dysplastic" appearance, which may signal impending duct loss. Focal duct loss can also occur as part of the spectrum of acute cellular rejection. However, cases in which features of cellular rejection are associated with more extensive duct loss ( affecting more than 50% of portal tracts ) should probably be regarded as having progressed to chronic rejection.
(3) Venous Endothelial Inflammation ("endotheliitis", "endothelialitis") This affects portal and hepatic venular endothelium. In early or mild cases there is lymphoid attachment to the luminal surface of endothelium. In more advanced or severe cases there is subendothelial infiltration associated with lifting and sometimes disruption of the endothelium. Some cases may be associated with veno-occlusive lesions .
Other Changes in Acute Rejection
Arterial lesions are rarely seen in acute rejection. When present they have been regarded as a sign of severe damage [7, 10].
Although inflammatory infiltration in acute rejection is predominantly portal, minor degrees of inflammation may also be present in the liver parenchyma. In addition to the periportal spillover described above, there are sometimes small areas of inflammation in perivenular regions. These are often associated with inflammatory lesions in hepatic venules and may be accompanied by small foci of perivenular dropout. A more diffuse pattern of sinusoidal inflammation is rarely seen and could represent "sinusoidal endotheliitis".
Other parenchymal changes frequently seen in association with acute rejection include cholestasis, ballooning, acidophil body formation and focal necrosis. These tend to be mainly present towards the acinar periphery and , in the early post-transplant period, may be related to "preservation injury", rather than rejection itself 
During the past 5 years, the author has been a member of a *European Liver Transplant Histopathologists' Group which has met regularly as part of the European Multicentre Study of the immunosuppressive agent FK506. The clinical results of this trial have been published recently . The group has formulated a consensus document on the diagnosis and grading of acute rejection, the main points of which are summarised in Table 3. Essentially this involves grading the three main diagnostic features of acute rejection in much the same way as already described for the Birmingham System and then combining these to produce an overall rejection grade. As an addendum, the grading scheme incorporates other features which have been suggested by others to be of adverse prognostic significance. The possible relevance of these is discussed later.
A modified version of the Birmingham System has recently been proposed by the group from the Royal Free Hospital (RFH) in London, U.K. . Using stepwise logistic regression analysis of histological features observed in 106 initial biopsies obtained during the first week post-transplant, 4 features were found to be important determinants in the diagnosis of rejection. These included the 3 main features listed above together with portal tract eosinophilia. All 4 feature s were graded on a scale of 0-3, producing a final rejection score between 0 and 12. This was then converted to a rejection grade as follows : 0-1 = no rejection, 2- 3 = consistent with rejection (not diagnostic), 4-6 = mild rejection, 7-9 = moderate rejection, 10-12 = severe rejection. Clinical correlation and prognostic value of the RFH grading system have not yet been assessed. However this study did show an excellent correlation between the RFH grade of rejection and 2 existing U.S. grading systems [7, 10]. This observation is perhaps surprising, especially with regard to severe rejection for which different diagnostic criteria are used in the 2 U.S. grading systems.