Performance Assessment Techniques of Liver Fibrosis Non-invasive Diagnostic Methods in an Approach without a 'Gold' Standard: Saying Goodbye to Liver Biopsy?

Autor: A. Yu. Ishchenko, M. Yu. Galushko, S. A. Pozdniakov, I. G. Bakulin
Rok vydání: 2020
Předmět:
Zdroj: Experimental and Clinical Gastroenterology. 1:4-11
ISSN: 1682-8658
DOI: 10.31146/1682-8658-ecg-170-10-4-11
Popis: Aims: 1. To evaluate the applicability of The Expert Assessment (EA) based on the algorithm developed by the authors in performance assessment of transient elastography by M-probe (TE-M) and XL-probe (TE-XL), Fibrotest® (FT), shear wave elastography (SWE) and liver biopsy (LB) in fibrosis staging in chronic hepatitis C (CHC) patients in an approach without a «gold» standard; 2. To evaluate the quality of the proposed model (EA efficacy) in comparison with the results of the Rasch Model (MR) being an extension of the Latent Variable Analysis.Material and methods. Using the five studied diagnostic methods liver fibrosis staging was performed in chronic hepatitis C patients. Based on the algorithm developed EA was applied to each patient of the cohort and as a result fibrosis stage was established. That fibrosis stage considered a virtual reference. Sensitivity and specificity estimation as well as AUC calculation for studied methods was performed against the reference standard. The quality of the model (EA efficacy) estimated in the Rasch Model (MR) being an extension of the Latent Variable Analysis.Results. 99 patients with reliable results of each of the five diagnostic methods were included into the study. The main characteristics of the cohort were: 59.6% individuals of male gender, average age of 37 years (21–63), median BMI of 25.8 kg / m2. As per the results of the EA in detecting of mild fibrosis (F1), the values of AUC, sensitivity and specificity were, respectively: 0,972; 94,3; 91,3 for TE-M; 0,964; 90,6; 91,3 for TE-XL; 0,806; 81,1; 73,9 for FT; 0,907; 88,7; 71,7 for SWE; 0,832; 92,5; 37,0 for LB. In detecting of moderate fibrosis (F2), the appropriate values were: 0,981; 93,8; 94,0 for TE-M; 0,967; 90,6; 97,0 for TE-XL; 0,873; 75,0; 80,6 for FT; 0,957; 84,4; 91,0 for SWE; 0,937; 90,6; 94,0 for LB. In detecting of significant fibrosis (F3), the corresponding values were: 0,994; 95,5; 96,1 for TE-M; 0,987; 90,9; 97,4 for TE-XL; 0,870; 63,6; 84,4 for FT; 0,961; 86,4; 97,4 for SWE; 0,990; 90,9; 98,7 for LB. In detecting cirrhosis (F4), the appropriate values were: 0,995; 92,3; 97,7 for TE-M; 0,994; 100,0; 98,8 for TE-XL; 0,874; 38,5; 93,0 for FT; 0,996; 92,3; 100,0 for SWE; 0,964; 69,2; 100,0 for LB. The EA was shown to have high degree of consistency (correlation coefficient = 0,923; p Conclusion. The EA was shown as a highly informative approach, applicable for performance assessment of the diagnostic methods for liver fibrosis staging without using LB as the “gold” standard. Non-invasive methods have demonstrated higher diagnostic characteristics with changing the standard from the traditional (LB) to virtual (EA). The high degree of consistency of the results of EA and MR confirm the high quality of the algorithm developed by the authors.
Databáze: OpenAIRE
Externí odkaz: