An algorithm for the grading of activity in chronic hepatitis C

Abstract

Histological activity reflects the global assessment of basic necroinflammatory lesions and is a criterion of major importance in chronic hepatitis C. The aim of this study was to propose and test the accuracy of a simple algorithm that generates a single activity score based on basic pathological features. A panel of 10 pathologists reviewed 363 chronic hepatitis C liver biopsies and graded the activity of hepatitis according to their own experience (reference activity). Then, a consensual algorithm on the grading of activity was established by the 10 experts in a panel discussion. Finally, stepwise discriminant analysis was performed to define which basic features had been intuitively used in the reference activity (statistical activity). To test the accuracy of the algorithm, concordance between the activity defined by the algorithm and the reference activity was assessed. It was compared with concordance between the activity defined by the statistical model and the reference activity. The algorithm proposed by the panel for the grading of activity included piecemeal necrosis and lobular necrosis. Concordance between reference activity and activity defined by the algorithm was substantial (305 cases, 84%, kappa = .75). Discriminant analysis showed that piecemeal necrosis, lobular necrosis, and portal inflammation were independently used to grade the activity. Concordance between reference activity and activity defined by the statistical model was substantial (300 cases, 83%, kappa = .73), virtually identical to the concordance between reference activity and activity defined by algorithm. This study proposes a simple algorithm for the grading of activity in chronic hepatitis. Its accuracy is as high as that obtained using a statistical approach.