Introduction
Liver elastography is one of the basic diagnostic tools in hepatology practice. The most widely used and most studied tool in the world is transient elastography (Fibroscan), however, this system is unevenly available in Slovakia. Shear-wave elastography is a more accessible method because it is built into standard ultrasound systems. However, the problem of shear-wave elastography remains the interpretation of its results, since the threshold values for significant fibrosis (SF) and advanced fibrosis (AF) are unique for each system.
Aims & Methods
To compare the results of transient elastography (TE) and point shear-wave elastography (pSWE) in a cohort of patients examined simultaneously with both systems and to determine the clinical benefit of pSWE in practice.
We examined all indicated patients over a 15-month period using both TE and pSWE (Hitachi Arietta V70). We recorded liver stiffness values in kPa, and baseline demographic and laboratory parameters. We used TE as the gold standard, with a threshold of >10 kPa for SF and >15 kPa for AF. We evaluated the accuracy of pSWE in predicting the detection of SF and AF. We used three models: the first one containing only pSWE values, and two logit models using pSWE together with common demographic and laboratory parameters. We identified the area under the curve (AUROC) and threshold values for each model and evaluated the probability of SF and AF according to these values.
Results
We examined 500 patients with a median age of 60 (IQR 48-69), male/female 214/286, with a SF rate of 30.8% and AF of 18.4%. The stiffness values according to TE and pSWE showed a highly significant correlation (rho=0.593, P<0.001). We found the accuracy of pSWE alone in predicting SF (AUROC=0.812), and AF (AUROC=0.839) with threshold values of SF>6.34 and AF>8.45 kPa. For values of pSWE<6.34, 6.34-8.45, and >8.45 kPa, the probability of SF was 11.1, 40.9, and 65.4% and AF was 5.0, 14.77, and 48.9%. We created two probability models (for P, range 0 to 1): ASTEL1 (N=369) that included pSWE, GGT, and platelets (AUROC SF=0.856, AUROC AF=0.908), and ASTEL2 (N=467) using pSWE, age, sex, AST/ALT, GGT (AUROC SF=0.857, AUROC AF=0.901). We set a threshold of P<0.33 to exclude SF and P>0.66 to confirm AF. At P<0.33, P=0.33-0.66, and P>0.66, SF was present in 15.6, 62, and 94.2% (ASTEL1) and 13.92, 50.7, and 86.8% (ASTEL2). At P<0.33, P=0.33-0.66 and P>0.66, AF was present in 5.6, 30, and 79.7% (ASTEL1) and in 4.4, 28, 73.68% (ASTEL2).
Conclusion
pSWE alone has an excellent negative predictive value (<6.34 kPa, 88.9-95%) in the diagnosis of SF and AF. Higher accuracy of positive prediction of SF (86.8-94.2%) and AF (73.7-79.7%) can be achieved by using one of two probability models (for P>0.66) combining SWE values with common demographic and laboratory parameters.
