EASL Clinical Practice Guidelines

Diagnostic work-up and diagnostic criteria

The diagnosis of AIH is usually based on the presence of the typical phenotype of the disease along with the exclusion of other causes of chronic liver diseases (Table 2, Table 4). The diagnostic criteria for AIH and a diagnostic scoring system have been codified by a group of experts in the IAIHG in 1993 [26], revised in 1999 [27] and more recently proposed in a simplified manner for routine clinical use (Table 6) [28].

Obvious features raise the suspicion of AIH and the application of published criteria (Table 5, Table 6) allows a ready diagnosis [26]. Unfortunately, in a considerable proportion of cases, the diagnosis is challenging and referral to hepatologists with specific clinical expertise in AIH may be warranted. In patients with an insidious onset and gradual progression without apparent symptoms, the diagnosis relies predominantly on laboratory findings. Therefore, the diagnostic work-up rests on such central elements as circulating non-organ specific autoantibodies associated with polyclonal hypergammaglobulinemia and typical or compatible histology in the absence of viral hepatitis markers. Histology is also essential in making the diagnosis.

Table 5
Summary of the criteria for the diagnosis of autoimmune hepatitis, on which the 1999 International Autoimmune Hepatitis Group (IAIHG) diagnostic score was based [27].

Laboratory findings

A predominantly hepatitic pattern, with bilirubin concentrations and aminotransferases ranging from just above the upper limits of normal to more than 50 times these levels, with usually normal or only moderately elevated cholestatic enzymes, is the typical biochemical profile of the disease [[4], [26], [27], [28]]. However, the degree of ALT elevation does not reliably reflect severity of AIH at the histological level. Of interest, recent studies have shown that along with the elevations of aminotransferases, γ-GT levels but not ALP can also be increased in AIH and furthermore, might be used as independent predictor of treatment outcome [[29], [45]]. In keeping with the fluctuating nature of the disease, the aminotransferases and γ-GT levels may even spontaneously normalise (spontaneous biochemical remission), despite histological evidence of persisting inflammatory activity, sometimes even severe inflammation. Such spontaneous apparent biochemical remissions are a critical issue that may sometimes result in delay and/or underestimation of the diagnosis, since reappearance of clinical disease may only become obvious several months or years later, or may even be completely asymptomatic. This disease behaviour may sometimes explain the presence of already established cirrhosis in almost one third of patients at the time of initial diagnosis.

Increased serum γ-globulin or IgG levels are found in approximately 85% of patients with AIH even in the absence of cirrhosis [[29], [104], [105]]. This prevalence tends to be lower in patients with an acute onset of the disease, in which a higher proportion of patients (25% to 39%) with normal IgG levels has been reported [[106], [107]]. The presence of high IgG levels is a very distinctive feature (IgA and IgM levels are usually normal) [28]. Increased IgA or IgM levels suggest different diseases such as alcoholic steatohepatitis and PBC, respectively.

It is important to underline that the range within which γ-globulins and IgGs are considered normal is wide. This may explain why a proportion of patients may show apparently “normal” IgG levels at diagnosis. Many, if not most of these patients have IgG levels in the upper range of normal, and show a marked fall upon initiation of therapy, sometimes even to levels below the normal range. These patients have a relative increase of their IgG levels considering their very low natural IgG levels but are still within the statistical normal range hampering initial diagnosis. The drop in IgG levels observed during treatment seems to confirm this hypothesis. Indeed, the level of immunoglobulins is an important and useful marker in monitoring the response to treatment and the achievement of remission. Reaching normal levels of immunoglobulins has been shown to correlate well with the improvement of inflammatory activity, even if sometimes a mild inflammatory activity (hepatitis activity index (HAI) 5–6) may coexist with normal IgG levels [108]. Normalisation of both transaminase levels and IgG levels has therefore been agreed upon as diagnostic marker of full biochemical remission [34].

The absence of viral markers is one of the four elements included in the simplified diagnostic criteria for AIH [28], but in countries with a high prevalence of viral hepatitis co-existence of AIH and viral hepatitis may exist [[109], [110], [111]]. In these cases the diagnosis of AIH may be overlooked and AIH could remain untreated, if absence of viral hepatitis is considered a prerequisite for making the diagnosis AIH. Usually AIH has a more aggressive course and more severe prognosis than viral hepatitis (either B or C) and a careful evaluation of the liver biopsy along with liver autoimmune serology testing can help in identifying the co-existence of a double mechanism of liver damage. With the advent of interferon-free regimens for the treatment of HCV infection, the possibility of treating both AIH and viral hepatitis has become much easier, and in milder cases HCV infection should be treated first and then liver disease reassessed.

Autoantibodies

Autoantibodies are the hallmark of AIH and represent an important part of the diagnostic work-up. Indirect immunofluorescence (IFL) is the preferable and main technique for routine autoantibody testing [112] for all autoantibodies except anti-SLA/LP antibodies (Fig. 2). It should be performed on freshly frozen rodent substrate that usually includes kidney, liver and stomach. This combination allows detection of ANA, SMA, anti-LKM1, as well as the rare antibodies anti-LC1 and anti-LKM3, if anti-LKM1 is absent. At the same time, AMA are also reliably detected by initial IFL screening and can thus help detect co-existent or variant forms of AIH-PBC. Positive sera should be titrated up to extinction. In adults, significant titers are ⩾1:40 dilution by IFL. In children, titers of 1:20 for ANA or SMA and 1:10 for anti-LKM1 are already strongly supportive of the diagnosis of AIH when used in combination with other laboratory and clinical features suggestive of the disease [54]. Other immunochemical techniques like ELISA or immunoblotting are available for the search of autoantibodies such as anti-LKM1, anti-LKM3, anti-LC1 and the only diagnostic tests for anti-SLA/LP, whose exact target antigens have been identified on a molecular level and are used in solid phase assays [[4], [112], [113]].


Fig. 2
A case based algorithm for patients with a suspicion of autoimmune hepatitis or drug-induced liver injury (DILI) using a response guided approach. *Test also for elevated IgG-levels. **These antibodies are highly specific for the diagnosis PBC.

View Large Image | View Hi-Res Image | Download PowerPoint Slide

ANA and SMA are markers of AIH-1, which account for about 75% of patients [[8], [11], [29]], but are not disease specific and show a wide range of heterogeneity in terms of antigenic specificity, together with a broad spectrum of titers. The fluorescence pattern of ANA in AIH is usually homogeneous using Hep2 cells but speckled pattern is not infrequent. The antibody is found in 43% of AIH-1 patients [29], and is associated with a variety of antigenic specificities including histones, double-stranded DNA (15%), chromatin and ribonucleoprotein complexes. However, no single pattern or combination is pathognomonic of AIH, whereas investigation for different staining of ANA patterns seems to have no practical clinical implications and diagnostic relevance in routine clinical practice and, therefore, the use of Hep2 cells at the screening stage of AIH is not recommended [[112], [113], [114]]. SMA reacts to several cytoskeletal elements including F-actin with a reported prevalence of anti-actin antibodies in 41% of patients. When kidney sections are utilised as a substrate for IFL SMAvg (vessel/glomeruli) and SMAvgt (vessel/glomeruli/tubules) patterns can be identified, which are frequently associated with, but not pathognomonic of, AIH. They correlate with F-actin antigenicity [112]. In the diagnostic work-up for AIH, SMA/anti-actin antibody testing is appropriate and may also be done by ELISA [[114], [115]]. However, IFL remains superior to ELISA and provides the best specificity/sensitivity compromise for testing for SMA. In fact, actin is not the only target antigen of AIH-specific SMA reactivity and thus ELISA can miss the diagnosis in about 20% of cases [[4], [112], [113], [116], [117], [118]]. ANA and SMA re-activities frequently coexist in the same serum and this improves the strength of the diagnosis.

Anti-LKM1 and/or anti-LC1 are the serologic markers of AIH-2. The two antibodies often coexist and in a series of 38 AIH-2 patients, the reported prevalence was 66% for anti-LKM1 and 53% for anti-LC1, respectively [29]. Unlike the antigen heterogeneity seen for ANA and SMA the major target autoantigen of anti-LKM1 has been clearly identified as the cytochrome P4502D6 (CYP2D6) and the formiminotransferase cyclodeaminase (FTCD) for anti-LC1. Despite a well characterized target antigen, neither anti-LKM1 nor anti-LC1 are disease specific, as they have been described in a small proportion (5–10%) of adult and paediatric patients with chronic HCV infection [[87], [88], [119], [120], [121], [122], [123]]. The presence of homology sequences between CYP2D6 and HCV proteins is the basis for the appearance of anti-LKM1 antibody in HCV patients who are genetically susceptible for AIH (mainly DRB1∗07 positive) through a mechanism of molecular mimicry [124]. Homologies of the target antigen of anti-LKM1 antibodies with other viral proteins have also been observed [5]. Also, anti-LKM1 antibodies have been described in a liver transplant recipient transplanted for Wilsońs disease following rejection episodes [125].

Anti-SLA/LP is the only disease specific autoantibody and therefore it has high diagnostic value. The target antigen has been identified as a synthase (S) converting O-phosphoseryl-tRNA (Sep) to selenocysteinyl-tRNA (Sec), whose terminological correct label is SepSecS [[126], [127]]. This has led to the development of reliable commercial assays for anti-SLA/LP detection (ELISA and dot-blot) [[126], [128]]. Anti-SLA/LP is detected in approximately 30% of patients with AIH more commonly associated with conventional autoantibodies, and is often associated with anti-Ro52 antibodies [[128], [129], [130], [131], [132]], but sometimes it is the only autoantibody reactivity detectable. Its presence may identify patients with more severe disease and worse outcome [[49], [50], [133], [134]] though these prognostic associations are controversial [[129], [132]].

Further autoantibody testing may be helpful, in particular for those patients testing initially negative in the above assays. Antineutrophil cytoplasmic antibodies (ANCA) are detected using ethanol-fixed human neutrophils with serum diluted 1:20. Atypical pANCA antibodies, originally considered specific of PSC and inflammatory bowel disease, are also frequently present in patients with AIH-1 [[135], [136]]. Recent evidence indicates that the target antigen is located in the nuclear membrane and for this reason some authors describe these antibodies as perinuclear anti-neutrophil nuclear antibodies (p-ANNA) [[137], [138]]. Their positivity can be an additional element used towards the diagnosis of AIH, particularly if other autoantibodies are negative [[27], [112]]. AMA, the specific serologic marker of PBC diagnosis, can be occasionally detected (8–12%) [[139], [140]] in patients with the classical phenotype of AIH without any other evidence of PBC, and may hint at co-existent or underlying PBC. Nonetheless, these patients should be classified and treated according to their clinical phenotype.

Autoimmune serology remains the Achilles’ heel in the diagnostic work-up for AIH. In fact IFL using rodent tissue, which has been indicated by the committee for autoimmune serology of the IAIHG as the best technique for the detection of autoantibodies is time-consuming, requires experienced technicians and is insufficiently standardised. Indeed, in real life the development of in-house validated sections for IFL does not seem to be feasible. Commercial substrates are also available; their quality however is variable. These are treated with fixatives in order to lengthen their shelf life, but this also causes enhanced background staining which can potentially cause difficulties in the interpretation of fluorescence patterns.

Therefore, methods other than IFL, like ELISA, are gaining popularity. This shift has been supported by the introduction of assays based on recombinant/purified target antigens (CYP2D6, FTCD, SLA/LP, and F-actin). However, the use of ELISA as the sole primary screening test for AIH-related autoantibodies is inappropriate because there is no useful combination of molecular specificities for a dependable detection of ANA and SMA, while the results are interchangeable with IFL for those autoantibodies (anti-LKM1, anti-LC1) whose target antigen has been identified at the molecular level [141]. Fig. 2 provides an algorithm for autoantibody testing in AIH.

Autoantibody titers and specificity may vary during the course of the disease, and seronegative individuals at diagnosis may express the conventional autoantibodies later in the disease course [[142], [143]]; in fact, repeated testing may allow autoantibody detection and, thus, correct disease diagnosis and classification [[27], [112], [113]]. In adults, autoantibody titers correlate only roughly with disease activity, clinical course and treatment response [144] and, therefore, they do not need to be monitored regularly unless a significant change in the clinical phenotype does appear. However, in paediatric patients, autoantibody titers may be useful biomarkers of disease activity and can be used to monitor treatment response [145]. In particular anti-LC1 antibodies have been shown to correlate well with disease activity showing a significant decrease in titer (>50%) or disappearance during remission and flare up during relapse [146].

The detection of autoantibodies plays a pivotal role in the diagnosis of AIH. Laboratory personnel and clinicians need to increase their expertise with disease expression and the interpretation of liver autoimmune serology in order to derive maximal benefits for patients. Tests must be ordered specifically on the basis of reliable clinical data and test results must not be interpreted outside the specific clinical context. Only then, can sensible evidence-based decisions be made, and the potential of serological work-up be exploited to the benefit of the patient. Finally, complete work-up for autoimmune serology is not available in all laboratories; it is important to identify laboratories, which are able to fully characterize patients’ sera, and patient sera should be sent to such reference laboratories for full evaluation especially in cases of diagnostic uncertainty.

Histology

Liver biopsy is considered a prerequisite for the diagnosis of AIH [[26], [27], [28]]. Apart from diagnosis, it is used to guide treatment decisions and should be performed before starting treatment, provided there are no contraindications [[27], [28]]. When severe coagulopathy is present the transjugular approach can be used, in particular, in acute/fulminant onset of the disease. Alternatively, biopsy under visual control by mini-laparoscopy has also been shown to be safe [147] even in cases of advanced coagulopathy [148], and may yield additional information [[147], [149]]. Interface hepatitis (hepatitis at the portal-parenchymal interface) with dense plasma cell-rich lymphoplasmocytic infiltrates, hepatocellular rosette formation, emperipolesis (active penetration by one cell into and through a larger cell) and hepatocyte swelling and/or pycnotic necrosis are the typical hallmarks of AIH [[28], [150], [151]]. Plasma cells are typically abundant at the interface and throughout the lobule, but their paucity in the inflammatory infiltrate (34% of cases) does not preclude the diagnosis [[150], [152], [153]]. However, there is no morphological feature that is pathognomonic of AIH. Interface hepatitis is not disease specific and patients with drug-related, viral or immune-mediated disease may show similar features.

Panlobular hepatitis, bridging necrosis and massive necrosis all of which are signs of severe inflammatory activity, are present less commonly but are part of the histological spectrum [27] and may occur in acute disease onset. The characteristic histological pattern is panacinar hepatitis (parenchymal collapse) especially in biopsies performed during an acute onset and closely resembles drug-induced hepatitis [[154], [155]]. Alternatively pericentral (Rappaport zone 3) necrosis may be present, which also resembles acute toxic injury [[107], [156], [157]]. These histological lesions have recently been proposed by the US NIH Acute Liver Failure Study Group as a set of diagnostic criteria for AIH presented as acute liver failure [37]. Additional features in cases of acute liver failure due to AIH include the presence of portal lymphoid follicles, a plasma cell-enriched inflammatory infiltrate and central perivenulitis [[37], [107]]. Transition from pericentral hepatitis to interface hepatitis has been demonstrated in sequential biopsies from patients with acute disease onset [158]. These observations suggest that the perivenular pattern of injury may be an early histological manifestation of AIH that is missed in biopsies obtained later in the course of the disease. Other lesions such as granulomata, cholangitis, steatosis or steatohepatitis can be seen, but if prominent they reduce the probability of the diagnosis of AIH. An inflammatory lymphocytic infiltrate of bile ducts has been described in 10% of cases but such individuals usually lack clinical, serological and immunological features of PBC, and they respond to corticosteroid therapy as patients with classical AIH [[63], [159]]. At the time of diagnosis different stages of fibrosis are present and about one third of patients already display established cirrhosis [[150], [151], [160]]. Macroscopic assessment by mini-laparoscopy increases the detection rate of cirrhosis by up to one third, as the macronodular nature of AIH cirrhosis may lead to false negative biopsy results missing the fibrous septa between regenerative nodules [[147], [161]]. Of interest, the histological features of necroinflammatory activity and severity of AIH are often not in parallel with the biochemical activity of the disease [[27], [28], [34], [40]]. Liver biopsy, therefore, provides information on prognosis and management as for instance the presence of cirrhosis may influence the choice and dose of the immunosuppressive agents prescribed while suggesting the need for regular screening for complications, such as oesophageal varices and HCC. It is highly recommended to have the histology reviewed by an experienced liver histopathologist who should discuss the most difficult cases with the clinician. The pathologist should weigh the inflammatory activity with the aid of the HAI score in order to give a quantitative evaluation of the inflammatory process to be monitored during treatment and follow-up. Despite the growing interest for non-invasive methods for the assessment of fibrosis and inflammation, most studies with these techniques have been performed in the field of viral hepatitis C and very few data are available on AIH. The limits for the clinical use of these methods in AIH, in particular Fibroscan, are related to the interference of necroinflammatory activity in the florid phase of the disease and to the overlap of adjacent stages of fibrosis [162]. A recent proposed non-invasive diagnostic score to predict inflammatory activity and severity of fibrosis based on routine laboratory parameters in AIH provides a useful tool for monitoring disease activity during treatment but cannot at present substitute the need of a biopsy, particularly at diagnosis [163].

Diagnostic scoring criteria

A comprehensive scoring system which grades every clinical, laboratory and histological feature of AIH, including response to corticosteroid treatment, has been published in 1999 by the IAIHG [27]. This scoring system was initially developed to define homogeneous cohorts of AIH patients for clinical trials rather than diagnosing AIH in individual patients. This scoring system has been validated in several papers [[164], [165], [166]] and, although developed as a research tool to provide comparability among populations in clinical trials, it has been widely used in clinical practice in assessing patients with few or atypical features of the disease not readily captured by the descriptive criteria [167]. The typical features, on which the diagnostic score is based, are summarized in Table 5. The drawbacks of this scoring system as a clinical tool are its complexity and failure to consistently distinguish AIH from cholestatic syndromes. In 2008, a simplified scoring system designed for every day clinical practice was proposed by the IAIHG [28]. It is based on four parameters: presence and titer of autoantibodies detected by IFL or ELISA (for anti-SLA/LP), serum IgG concentration, presence of typical or compatible histology, and absence of viral hepatitis markers (Table 6). Compared with the original revised score system the simplified score has somehow lower sensitivity (95% vs. 100%) but higher specificity (90% vs. 73%) and accuracy (92% vs. 82%) [[55], [166], [168], [169]] and, according to some studies, seems to work well in patients with AIH-PBC variant [[168], [170]]. It does not grade response to corticosteroid therapy [169] whose inclusion has been advocated as an additional criterion, but is primarily meant to serve as a guide for the initiation of treatment, at which time point information on response can naturally not be available. The simplified scoring system is useful in excluding AIH in patients with other conditions and concurrent immune features [[166], [170]], but it is more likely to result in the exclusion of atypical cases [[166], [169], [171]]. Prospective evaluation of these criteria is required to corroborate these observations.

Table 6
Simplified diagnostic criteria of the International Autoimmune Hepatitis Group [28].

Definite autoimmune hepatitis: ⩾7; Probable autoimmune hepatitis: ⩾6. ∗Addition of points achieved for all autoantibodies (maximum, two points). Typical liver histology for autoimmune hepatitis = each of the following features had to be present namely, interface hepatitis, lymphocytic/lymphoplasmacytic infiltrates in portal tracts and extending into the lobule, emperipolesis (active penetration by one cell into and through a larger cell), and hepatic rosette formation. Compatible liver histology for autoimmune hepatitis = chronic hepatitis with lymphocytic infiltration without all the features considered typical. Atypical = showing signs of another diagnosis, like steatohepatitis.

In conclusion, the simplified criteria are user-friendly and a good tool for daily clinical practice but without a diagnostic “gold standard” the clinicians must regard any diagnostic score only as an aid to diagnosis of AIH [172] and the criteria should be used alongside clinical judgement. In this context, a subgroup of patients especially those with acute or fulminant onset of AIH may be missed by standard diagnostic criteria and therefore require special attention. Acute or fulminant AIH is characterized by an abrupt onset of symptoms and frequently with acute liver failure. The challenge in the diagnosis is related to the lack of a widely accepted definition [173] and of the phenotype characteristic of AIH. In fact, 25% to 39% of patients with acute onset AIH have normal levels of IgG [[106], [107]] while 9% to 17% test negative for circulating autoantibodies [[39], [106]]. The higher percentage of those patients with normal IgG in this setting can be explained by the short duration of the inflammatory process. A liver biopsy is essential although the classical lesions pathognomonic of AIH are frequently lacking and the most prominent lesion is the pericentral necrosis [[37], [156], [157], [158]]. So far, few data are available on the use of the published scoring systems in acute onset of AIH [[169], [174]]. In 70 patients with fulminant liver failure, the revised scoring system supported the diagnosis of AIH in 40% of cases, whereas only 24% were identified by the simplified scoring system [169]. In another series of 55 patients with acute/fulminant onset 91% had a score compatible with AIH with the revised scoring system vs. 40% with the simplified version [174]. The use of the diagnostic scoring systems in this setting of patients should be further evaluated prospectively.