Transient Elastography: Defining the optimal cutoff for estimating cirrhosis
S. Naimimohasses, O. El-Sheriff, H. Irish, B. Hynes, S. Norris, S. McKiernan
Hepatology Department, St. James Hospital, Dublin
International guidelines have moved towards recommending non-invasive measures for assessing hepatic fibrosis in a variety of different conditions. One of the most established methods with the largest supporting body of evidence is Transient Elastography.
To validate results obtained via use of a fibroscanner and to determine an optimal cut-off with the highest sensitivity and specificity for cirrhosis.
We performed a retrospective study of patients who underwent Transient Elastography with an ECHO-sens Fibroscanner with an M2 probe, since its introduction to the Hepatology Department at St. James Hospital from 2008 to 2015 and conducted a direct comparison with biopsies performed within the same year.
1394 measurements were performed over a 7-year period. Of those, 123 underwent a liver biopsy within 12 months of scanning. 68.3% (83) of patients had hepatitis C, 8.9% (11) had hepatitis B, 6.5% (8) were diagnosed with NASH, 3.3% (4) had autoimmune hepatitis and cholestatic liver disease. 5.7% (7) were diagnosed with cryptogenic cirrhosis and 2.4% (3) had haemochromatosis.
The ROC curve when all data sets were included demonstrated an AUC of 0.89. A score of greater than or equal to 13.5kPa provided an optimal cut off for estimating cirrhosis yielding a sensitivity of 87% and specificity of 82%. Only 50.4% (62) of measurements had a combination of 10 measurements with success rates of greater than or equal to 90% and an IQR: Mean Liver Stiffness ratio of ≤30%. Interestingly when ROC curves were adjusted for success rate, the AUC increased when a lower than standard threshold was included (AUC= 0.887 with success rates of ≥ 90% and 0.917 with success rates of ≥ 50% with a sensitivity of 84.2% and a specificity of 85.7% for a cut off of 13.5kPa).
This retrospective data analysis conducted demonstrated a good predictive value for identifying cirrhosis, independent of success rate. ICORN criteria recommend a fibroscan score cut off of ≥ 12.5kPa, which is supported by our data providing a sensitivity of 87% and a specificity of 81%, comparable with those obtained by high frequency ultrasound. It is recommended that all centers undergo regular validation of Fibroscan results obtained.