histological classification of chronic hepatitis, and several Histological activity reflects the global assessment of propositions have recently been made by different groups of basic necroinflammatory lesions and is a criterion of mapathologists. [4][5][6][7] A consensual proposition was to separately jor importance in chronic hepatitis C. The aim of this assess, using scoring systems, the degree of liver fibrosis and study was to propose and test the accuracy of a simple that of activity. 4,6 Liver fibrosis is simple to precisely define algorithm that generates a single activity score based and thus to reproducibly score. 8 In contrast, the grading of on basic pathological features. A panel of 10 pathologists activity, which integrates the different basic necroinflammareviewed 363 chronic hepatitis C liver biopsies and tory lesions, is more difficult to assess; morever, it is of imporgraded the activity of hepatitis according to their own tance in the decision of whether or not to treat patients in experience (reference activity). Then, a consensual algomany therapeutic trials. Although not proven, inflammation rithm on the grading of activity was established by the and liver cell necrosis are the hallmarks of active disease 10 experts in a panel discussion. Finally, stepwise disthat may predict evolution toward fibrosis and cirrhosis. 9 criminant analysis was performed to define which basic Several approaches have been proposed to assess histologifeatures had been intuitively used in the reference activcal activity. One of them is the semiquantitative scoring sysity (statistical activity). To test the accuracy of the algotem of Knodell et al., in which fibrosis and portal, periportal, rithm, concordance between the activity defined by the and lobular necrotic and inflammatory components are asalgorithm and the reference activity was assessed. It was sessed separately and their coding values added. 10 The global compared with concordance between the activity descore appears accurate because it varies over large ranges, fined by the statistical model and the reference activity.
but its value may be limited by poor reproducibility, because The algorithm proposed by the panel for the grading of each feature has its own observer variation. 8 Another possible activity included piecemeal necrosis and lobular necroapproach is to consider that periportal and intralobular nesis. Concordance between reference activity and activity croinflammatory lesions are related to the same pathogenic defined by the algorithm was substantial (305 cases, 84%, mechanism and that they must be globally assessed. As prek Γ
.75). Discriminant analysis showed that piecemeal viously reported, assessment of activity must be separated necrosis, lobular necrosis, and portal inflammation were from staging of fibrosis. 4 independently used to grade the activity. Concordance
The aim of this study was to propose a simple algorithm between reference activity and activity defined by the that generates a single score of necroinflammatory activity statistical model was substantial (300 cases, 83%, k Γ
.73), based on basic histological features. The creation of this algovirtually identical to the concordance between referrithm drew on the experience of a panel of pathologists, and ence activity and activity defined by algorithm. This its validation was performed by comparison with another study proposes a simple algorithm for the grading of method for establishing activity based on a statistical apactivity in chronic hepatitis. Its accuracy is as high as proach. that obtained using a statistical approach. (HEPATOLOGY 1996;24:289-293.)
PATIENTS AND METHODS
Liver Biopsies. A total of 363 percutaneous liver biopsies from pa-Liver biopsy is a gold standard in chronic hepatitis and is tients with chronic hepatitis C were included in this study. All pathe only investigation that can estimate the severity of tissue tients were positive for antibody to hepatitis C by second-generation damage, because clinicopathological correlations are poor in tests and had at least a 6-month increase in aminotransferase levels. chronic hepatitis C. However, the value of a liver biopsy de-All biopsies were from patients included in five French therapeutic pends on the possibility of assessing histological features ustrials on chronic hepatitis C, the results of which have already been ing a standardized reproducible classification.
published. [11][12][13][14][15] Biopsy was performed at the time of inclusion, when With the discovery of the hepatitis C virus and its worldpatients had not received any treatment. Liver biopsies, measuring more than 10 mm in length, were fixed, paraffin-embedded, and wide diffusion, pathologists have been faced with an increasstained with hematoxylin-eosin safran and Masson's trichrome, or ing number of liver biopsies showing chronic hepatitis. [1][2][3] Picrosirius red for collagen. Biopsies were anonymously coded. Pa-They have been tempted to discuss the terminology and the thologists were not aware of clinical and biochemical data, but only of the presence of chronic hepatitis C. Panel of Pathologists. A panel of 10 French senior pathologists experienced in the field of liver pathology was established. Patholo-The METAVIR group includes Pierre Bedossa (Le Kremlin-Bice Λtre), Paulette Bioulacgists had already worked together in the French METAVIR group, 8,16 Sage (Bordeaux), Patrice Callard (Paris), Miche le Chevallier (Lyon), Claude Degott (Clichy), and a consensus was thus reached on the definition and coding of Yves Deugnier (Rennes), Monique Fabre (Le Kremlin-Bice Λtre), Michel Reyne s (Villejuif), Jean-Jacques Voigt (Toulouse), Elie Serge Zafrani (Cre Β΄teil, France). the lesions. Intra-and interobserver variations in the assessment of From the Service d'Anatomie Pathologique, Ho Λpital de Bice Λtre, Le Kremlin-Bice Λtre, histological lesions in chronic hepatitis C had already been pub-