Impact of quality control accepted inter-laboratory variations on calculated Fibrotest/Actitest scores for the non-invasive biochemical assessment of liver fibrosis

Introduction: Non-invasive, i.e. serum-based assessment of liver fibrosis is still an important diagnostic challenge although multiple single and multiparametric panels of biomarkers have been suggested. Aim: Two approaches were followed to determine the diagnostic reliability of Fibrotest and Actitest, two commercially distributed non-invasive multiparametric tests for staging and grading of liver fibrosis. Methods: (i) Haptoglobin, ALT, gammaGT, alpha-2-macroglobulin, apolipoprotein A1 and bilirubin, required for calculation of respective scores, were determined in sera of 4 patients with histologically defined stages of fibrosis (F1-F4) and activities of fibrogenesis (A1-A3). Analytes were determined by 6 quality controlled external laboratories. Inter-laboratory variations of the calculated Fibrotest score for staging and Actitest score for grading (BioPredictive (TM)), and their error ratios referred to histologic results were calculated. (ii) The variability of respective Fibrotest/Actitest scores depending on 64 selected combinations of analytes within the accepted ranges of analyte-specific maximum/minimum limits given by the external quality control of the German Association of Clinical Chemistry and Laboratory Medicine (DGKL) was calculated. Results: (i) Fibrotest and Actitest scores were largely reproducible among the different laboratories. However, the error ratio was 77% for all results calculated by both, Fibrotest and Actitest when referred to histologic findings. (ii) Calculated scores of stages varied between F2 (9%), F3 (31%), F3-F4 (6%), and F4 (54%) (Fibrotest), and A1/A2 (48%), A2 (9%), A2-A3 (5%), and A3 (38%) for grades of fibrogenic activity. Conclusion: Despite reproducibility of Fibro- and Actitest scores among the six laboratories, large scale investigation displayed high levels of variability depending on inter-laboratory differences that were still in a quality controlled, analytically acceptable range. Calculated scores coincided with histologic findings in less than 25% of all cases. Thus. the diagnostic accuracy of these tests must be considered as low, if histology is accepted as reference standard. (C) 2009 Published by Elsevier B.V.