Additional tests for hepatitis A, B and C were all negative. No evidence of Epstein–Barr virus or cytomegalovirus was found. Negative serum copper tests and normal ceruloplasmin levels made Wilson's disease unlikely. 1-Antitrypsin levels were normal. IgG levels were increased (35.3 g/l) with normal IgA levels. Liver function was diminished with increased clotting time (INR 1.4, APTT 33.4 s), but albumin and ammonia levels were normal. There were no signs of anaemia (Hb 8.1 mmol/l). Sodium, calcium and phosphate levels were normal. Abdominal ultrasound showed a normal liver size with homogenous aspect and no other signs of abdominal abnormalities.

The girl was referred to the university hospital for a liver biopsy and start of treatment. The biopsy was very suggestive of autoimmune hepatitis, showing chronic, strongly active hepatitis with intra- and periportal infiltrates consisting mainly of plasma cells (figure 1). In addition, signs of parenchymal deterioration were seen as reflected by Councilman bodies and ballooning hepatocytes. Minimal periportal fibrosis was present. There were no signs of bile duct involvement. Copper staining was negative.

Tests for autoimmune hepatitis-associated autoantibodies were performed. Antinuclear antibodies (ANA), antiliver kidney microsomal type 1 (anti-LKM-1) antibodies, antiliver cytosol type 1 (anti-LC-1) and soluble liver antigen (SLA) antibodies were negative. Antismooth muscle antibodies (SMA) were not tested at diagnosis, but 6 months later, after therapy was initiated, they were weakly positive. Antineutrophilic cytoplasmic antibodies (ANCA) were negative. MRI-cholangiography showed no sign of sclerosing cholangitis.

Interestingly, antibody tests revealed high levels of antibodies against glutamic acid decarboxylase (GAD) and insulin, which are frequently present in patients with type 1 diabetes, reflecting the autoimmune processes underlying this disease. Insulinoma-associated tyrosine phosphatase (IA2) autoantibody, another autoantibody found in diabetes, was negative. These antibodies were not determined at diagnosis of diabetes mellitus 9 months earlier. Another interesting finding was the increased level of thyroid peroxidase antibodies (193 kU/l). Nonetheless, thyroid function tests were normal. Antiadrenal, antiendomysium and antitissue transglutaminase antibodies were negative. Additional human leucocyte antigen (HLA)-typing showed the following: DRB1*0301/DRB1*07, DQB1*02, DQA1*0201/DQA1*0501.

Two types of childhood autoimmune hepatitis are recognised by the IAIHG: type 1 (about 62% of cases), characterised by the presence of ANA and/or SMA antibodies and type 2 (about 38% of cases), characterised by the presence of anti-LKM-1 antibodies.⁶ The presence of anti-LC-1 has also been associated with autoimmune hepatitis type 2.⁸ Both types occur predominantly in females. Type 2 mainly occurs in children, whereas type 1 also occurs in adults. A third type of autoimmune hepatitis has been described which is characterised by the presence of SLA autoantibodies.⁹ Seronegative autoimmune hepatitis is rare in children and, except for a recent case report,¹⁰ has only been reported in the paediatric literature once in abstract form. In the present case, use of the scoring system means the diagnosis of autoimmune hepatitis is definite. However, whether it is type 1, type 2 or seronegative autoimmune hepatitis is unclear, since autoantibody tests at diagnosis were not complete and were performed according to a protocol standardised for adult testing, which may not be sensitive enough for children. The weak positivity for anti-SMA antibodies after 6 months of therapy indicates the presence of type 1 autoimmune hepatitis.

There are no randomised controlled treatment trials in children with autoimmune hepatitis. Therefore, drugs and treatment regiments are based on research in adults and reflect the preferences of individual centres. In adults ‘standard therapy’ is considered to be predniso(lo)ne combined with azathioprine.¹¹ About 80–85% of all patients with autoimmune hepatitis respond well to this treatment. The overall frequency of azathioprine-related side effects in patients with autoimmune hepatitis is 10%.¹² Side effects include cholestatic hepatitis, pancreatitis, nausea, emesis, rash, opportunistic infection, bone marrow suppression and, in the long-term, malignancy. Azathioprine treatment should not be started in patients with severe pretreatment cytopenia or known complete deficiency of thiopurine methyltransferase activity. In children, especially those with diabetes, decreasing the predniso(lo)ne dose is beneficial. Addition of azathioprine usually means the predniso(lo)ne dose can be lowered. Next to azathioprine, other immunosuppressive agents have been suggested and used in patients with autoimmune hepatitis, such as mycophenolate mofetil, tacrolimus and ciclosporin. However, results of the use of these drugs in the treatment of autoimmune hepatitis are restricted to small case series, mostly performed on adults.¹¹ Prognosis in children with autoimmune hepatitis who respond to immunosuppressive therapy is generally good, with most patients surviving long term with excellent quality of life on low-dose medication. However, the development of endstage liver disease requiring liver transplantation despite treatment has been reported 8–14 years after diagnosis in 8.5% of children with autoimmune hepatitis.⁷

Type 1 and type 2 autoimmune hepatitis have a high frequency of associated autoimmune disorders (type 1: 5–20%, type 2: 18–22%) and a family history of autoimmune disease (40–43%).^{7 13} Associated autoimmune disorders include thyroiditis, inflammatory bowel disease, vitiligo, autoimmune haemolytic anaemia, coeliac disease, nephrotic syndrome and type 1 diabetes. However, in these studies the co-occurrence of autoimmune hepatitis with diabetes mellitus in children is rare, comprising 4–6% of all autoimmune hepatitis cases. To our knowledge, our case is only the sixth child ever mentioned in the literature as having both type 1 diabetes and autoimmune hepatitis. Interestingly, in a study on patients with autoimmune hepatitis who were screened for the presence of diabetes-associated autoantibodies, 18.5% had insulin autoantibodies and one had GAD autoantibodies without clinical diabetes.¹⁴

The pathogenesis behind the co-existence of different autoimmune diseases is unclear. Although all autoimmune disease are characterised by increased autoantibodies, it is unlikely that they play a direct pathogenic role. The presence of autoantibodies merely reflects the (asymptomatic) phase that occurs after autoimmunity has been initiated, as has been demonstrated in type 1 diabetes.^{3 15 16} The trigger that initiates autoimmunity in the first place, however, is ill-defined. Both in type 1 diabetes and autoimmune hepatitis an underlying genetic predisposition has been suggested, because both diseases are associated with certain HLA, especially HLA DR3 and DR4.^{3 8} For example, in Northern Europe type 1 autoimmune hepatitis is associated with the presence of HLA DRB1*03,¹⁷ for which the girl in the present case was heterozygote. Interestingly, type 1 diabetes is also associated with HLA DRB1*03.^{3 4 15} However, the exact mechanisms involved in this possible common genetic association still need to be unravelled. The fact that multiple genes are associated with disease susceptibility suggests that dysregulation of multiple pathways of self-tolerance is needed for autoimmunity to occur, including dysregulation in central tolerance as in peripheral tolerance.³ It could very well be that different autoimmune diseases are induced by common dysregulations in pathways of self-tolerance. An individual with an autoimmune disease may therefore be more susceptible to develop another autoimmune disorder when triggered.

The co-occurrence of autoimmune hepatitis in type 1 diabetes is rare, while type 1 diabetes occurs frequently. Therefore, routine screening for autoimmune hepatitis-associated antibodies does not seem necessary. Nevertheless, alertness to possible symptoms of autoimmune hepatitis is indicated, especially because the clinical patterns of autoimmune hepatitis at onset of disease are various and can range from mild symptoms followed by jaundice to acute hepatic failure with encephalopathy and/or complications of portal hypertension at first presentation.⁶

In summary, we have presented the case of a 12-year-old girl with type 1 diabetes who developed autoimmune hepatitis. Apart from autoimmune hepatitis, patients with type 1 diabetes have a higher occurrence of other autoimmune diseases including autoimmune thyroiditis and coeliac disease. Therefore, there should be a high index of suspicion for other forms of autoimmunity in patients with type 1 diabetes.