Z Gastroenterol 2010; 48 - P5_24
DOI: 10.1055/s-0029-1246569

Improved diagnostic accuracy of transient elastography for cirrhosis using histology plus clinics as gold standard

S Mueller 1, G Millonig 2, F Stickel 3, T Longerich 4, P Schirmacher 5, HK Seitz 1
  • 1Zentrum für Alkoholforschung, Lebererkrankungen und Ernährung, Krankenhaus Salem und Universität Heidelberg, Heidelberg, Heidelberg
  • 2Zentrum für Alkoholforschung, Lebererkrankungen und Ernährung, Krankenhaus Salem und Universität Heidelberg, Heidelberg, Heidelberg
  • 3Institue of clinical Pharmacology, University of Bern,, Bern, Switzerland
  • 4Institut für Pathologie, Heidelberg, Heidelberg
  • 5Institut für Pathologie, Universität Heidelberg, Heidelberg

Aims: Measurement of liver stiffness (LS) by transient elastography (Fibroscan, FS) is a novel noninvasive approach to assess liver fibrosis with a high diagnostic accuracy. However, accuracy may be even higher since histology served as gold standard in all previous LS studies although its sample error may reach up to 30%. Consequently, we here determined the AUROC for FS to detect liver cirrhosis using a combination of histology and definite clinical signs of cirrhosis as obtained by ultrasound or gastroscopy in patients with alcoholic liver disease (ALD). Method: LS was measured in 90 patients with histologically confirmed ALD. A LS of >8kPa was considered as cut-off for F3/4 fibrosis. Patients with significant steatohepatitis (GOT>100 U/l) were excluded since inflammation increases LS irrespective of fibrosis. We finally compared FS-results to the combination of the histological fibrosis score (Kleiner) plus additional clinical information (macronodular surface or collateral circuits in ultrasound imaging, varices in upper GI endoscopy). Results: 77 patients were scored correctly by FS. 4 of 39 patients (10.2%) with histological F3/4 fibrosis had a LS <8 kPa but no clinical signs of liver cirrhosis. 9 of 51 patients (17.6%) with histological F0–2 fibrosis had a LS >8 kPa, 2 of them had definite clinical signs of liver cirrhosis. 2 more patients had an enlarged spleen (>12 cm) suggesting portal hypertension. Thus in 8 of 90 patients (8.8%) additional clinical information was able to resolute the divergence between histology and fibroscan. AUROC for the detection of F3/4 fibrosis by FS increased from 0.923 to 0.985 when histology in combination with clinical signs of cirrhosis was used as gold standard. Conclusion: FS results should be compared against a combination of histology and clinical information to minimize the effects of sampling error in liver biopsy. Using this as the new gold standard, FS reaches an AUROC of 0.98 in F3/F4 diagnosis.