Significance of histopathological features in differentiating autoimmune liver disease from nonautoimmune chronic liver disease in children

Objectives Autoimmune liver disease (AILD) requires a constellation of clinical, serological, biochemical, and histological findings for diagnosis. Liver biopsy forms the cornerstone for the definite diagnosis of AILD, despite histological features not being pathognomonic. Liver biopsies of AILD and nonautoimmune chronic liver disease (NACLD) were reviewed blindly to assess the role of typical histological findings in differentiating AILD from NACLD in a pediatric population. Participants and methods Twenty-five liver biopsies of AILD and 34 liver biopsies of NACLD were reviewed retrospectively without knowledge of the final diagnosis. Results The typical histology comprising all four features, interface hepatitis, portal lymphoplasmacytic infiltrate, rosette formation, and emperipolesis, was observed in 56% of AILD. Rosette formation and emperipolesis were associated significantly with the diagnosis of AILD. Rosette formation alone or in combination with emperipolesis or lymphoplasmacytic infiltrate had high specificity (96.2% each) but low sensitivity (68, 60, and 60%, respectively) for AILD. The diagnostic accuracy of typical histology comprising of a combination of at least three of four features, rosette formation, emperipolesis, and lymphoplasmacytic infiltrate, was 76.9%, with a positive predictive value of 93.3% and a negative predictive value of 70.2%. Conclusion Characteristic patterns of liver injury comprising typical histological features on liver biopsy may strongly suggest the diagnosis of AILD irrespective of other laboratory parameters in children. Rosette formation was the only independent significant histological factor to predict AILD. High specificity and predictability of typical histological features may be helpful in diagnosing seronegative AILD among cases of cryptogenic liver disease in the absence of other supportive findings. Eur J Gastroenterol Hepatol 25:333-337 (C) 2013 Wolters Kluwer Health vertical bar Lippincott Williams & Wilkins.